Protective Role of Arginase in a Mouse Model of Colitis¹

Despite extensive study, there is uncertainty about the regulation of inflammatory responses in human inflammatory bowel disease and in experimental colitis models. In such models, there is strong evidence for a protective role of cyclooxygenase-2 (1), TGF-β, and IL-10 (2), and a proinflammatory effect of IL-12 (3). However, in the case of NO, produced by the enzyme-inducible NO synthase (iNOS),⁴ there have been inconsistent results concerning its role in colitis. For example, iNOS inhibition has been reported to either reduce (4) or increase (5) the severity of trinitrobenzene sulfonic acid (TNBS) colitis, while iNOS deletion has been shown to exacerbate acetic acid (6) or TNBS (7) colitis, but to improve dextran sulfate sodium (DSS) colitis (8, 9).

Evidence is accumulating that the activity of the enzyme arginase, which competes with iNOS for the common substrate, l-arginine (l-Arg), has important biologic effects on host innate immune response (10, 11, 12) and epithelial cell functions (13, 14). The cytoplasmic type I arginase (15) and the mitochondrial type II arginase (16) catalyze the same conversion of l-Arg into ornithine, the latter being converted to polyamines by ornithine decarboxylase (ODC). Thus, arginase is an inhibitor of NO synthesis and can also modulate polyamine synthesis (17). Through these functions, arginase may have important biological effects in host mucosal immune response, which have been demonstrated in vitro (11, 18). However, the role of arginase has not been studied in the gastrointestinal mucosa in vivo. We hypothesized that arginase is an unrecognized regulator of mucosal inflammation and host-pathogen interactions in the gut.

Citrobacter rodentium is the rodent equivalent of enteropathogenic Escherichia coli (EPEC), which causes diarrhea in humans. C. rodentium colonizes the surface of colonic epithelial cells, resulting in signal transduction events, cytoskeletal rearrangements, formation of attaching and effacing lesions, epithelial hyperplasia, and a strong mucosal Th1 response (19, 20). Murine infection provides an ideal model, with reproducible histologic changes similar to human inflammatory bowel disease. The aim of this study was to investigate the roles of iNOS, arginase, and ODC in C. rodentium colitis. There was marked up-regulation of iNOS, arginase I, and ODC in the colon of infected mice. Both wild-type (WT) and iNOS^−/− mice supplemented with l-Arg had improvement in colitis, while disease was worsened when mice were treated with S-(2-boronoethyl)-l-cysteine (BEC), a specific arginase inhibitor (21), or α-difluoromethylornithine (DFMO), an ODC inhibitor. These findings suggest a protective role for arginase in the colon through the formation of polyamines.

A WT strain of C. rodentium (DBS100), provided by D. Schauer (Massachusetts Institute of Technology, Cambridge, MA), was used. Before infection, C. rodentium were grown in Luria broth overnight, washed, and resuspended in PBS, and bacteria concentration was determined by OD and confirmed by serial dilution and culture (18).

All studies were approved by the University of Maryland (Baltimore, MD) Institutional Animal Care and Use Committee. WT and iNOS^−/− C57BL/6 mice were obtained from The Jackson Laboratory (Bar Harbor, ME), and bred in our facility. Eight-week-old mice were gavaged with C. rodentium in 100 μl of PBS or PBS vehicle alone. Initial studies determined that an inoculum of 5 × 10⁸ bacteria/mouse was required for consistent induction of colitis. In some studies, mice were treated with 1% l-Arg (Sigma-Aldrich, St. Louis, MO), starting day 1 postinfection; 0.1% BEC (provided by J.-L.B.), starting day 3; or 2.5% DFMO (Ilex Oncology, San Antonio, TX), starting day 3, all at pH 7 in the drinking water. Daily weights were obtained for each mouse. Water consumption was measured and found to be similar between all experimental groups. Mice were sacrificed after 14 days or when moribund. Blood samples were obtained by intracardiac puncture, and colons were collected, weighed, and divided. Tissues snap frozen in liquid nitrogen were used for RNA studies, Western blotting, enzyme activity assays, and polyamine measurements; tissues fixed in 10% buffered formalin were used for histology and immunohistochemistry; fresh tissues were homogenized in PBS; and colonization levels were determined by serial dilution and culture.

Sections (6 μm) were cut from paraffin sections and stained with H&E. Tissues were examined in a blinded manner by a pathologist (C.B.D.). Acute (neutrophilic) and chronic (lymphocytic) inflammation, and epithelial regenerative changes were each scored on a 0–4 scale, and the sum was used as an index of histologic injury.

Immunohistochemistry was performed, as described (22, 23). Polyclonal Ab to bovine liver arginase I (1/200; Research Diagnostics, Flanders, NJ) or to murine iNOS (1/400; BD Transduction Laboratories, Lexington, KY) were used for 1.5 h at room temperature. Sections were washed and incubated with goat biotinylated anti-rabbit secondary Ab (1/200; Vector Laboratories, Burlingame, CA) for 15 min. The Vectastain Elite ABC kit (Vector Laboratories) was used as a peroxidase system. Color was developed using 3,3′-diaminobenzidine, and hematoxylin was used as a counterstain.

Frozen colon samples were homogenized in lysis buffer containing protease inhibitors (11). Protein concentrations of 14,000 × g supernatants were measured (DC Protein Assay kit; Bio-Rad, Hercules, CA), and 100 μg/lane was separated by SDS-PAGE and transferred onto Immobilon-P membranes (Millipore, Bedford, MA) by electroblotting. Membranes were blocked overnight at room temperature with PBS containing 0.1% Tween and 5% nonfat dry milk. Polyclonal Ab to arginase I (1/1000) or arginase II (1/1000; provided by R. Iyer and S. Cederbaum, University of California, Los Angeles, CA (11)) were used for 1 h at room temperature. Blots were washed and incubated with donkey anti-rabbit Ab conjugated to HRP (1/2000; Amersham Biosciences, Piscataway, NJ). Chemiluminescence was detected with SuperSignal West Pico chemiluminescent substrate (Pierce, Rockford, IL) and exposure to Hyperfilm ECL (Amersham Biosciences).

Protein supernatants were incubated with 0.1 μM l-[guanido-¹⁴C]Arg (NEN, Boston, MA) for 24 h at 37°C, and arginase activity assay was performed, as reported (10).

Frozen tissues were homogenized in a solution of 25 mM Tris-HCl (pH 7.4), 0.1 mM EDTA, and 2.5 mM DTT. Supernatants of 14,000 × g centrifugations were incubated at 37°C with 10 nmol of l-[1-¹⁴C]ornithine (NEN) for 24 h, and ODC activity was assayed, as reported (11). Bacterial ODC activity was measured after sonicating C. rodentium at log phase of growth. Polyamine concentrations were measured in tissues by HPLC, as described (24), and standardized to protein concentration. Data shown are the total concentration of the biogenic polyamines putrescine, spermidine, and spermine.

Colon samples were homogenized in 1 ml of TRIzol reagent (Invitrogen Life Technologies, Grand Island, NY), and total RNA was isolated. Reverse transcription, PCR cycle conditions, and primer sequences for iNOS, arginase I, arginase II, ODC, and β-actin were as described (11). PCR products were run on 2% agarose gels with 0.4 μg/ml ethidium bromide. Bands were visualized under UV light and photographed with a digital gel documentation system (EDAS 290; Kodak Digital Science, Rochester, NY). Real-time PCR was performed with a Bio-Rad iCycler iQ detection system for iNOS, arginase I, arginase II, and ODC, and with an MJ Research DNA Engine Opticon 2 (Waltham, MA) for IFN-γ, TNF-α, and IL-1. Fluorescein dye and SYBR green (Molecular Probes, Eugene, OR) were included in the PCR mix. For ODC, 3 pmol, and for iNOS, arginase I, arginase II, IFN-γ, TNF-α, IL-1, and β-actin, 7 pmol each of sense and antisense primers was used in each reaction. Sense and antisense primer sequences for the cytokines were: IFN-γ, 5′-GCCACGGCACAGTCATTGAA-3′ and 5′-CGCCTTGCTGTTGCTGAAGA-3′; TNF-α, 5′-CTGTGAAGGGAATGGGTGTT-3′ and 5′-GGTCACTGTCCCAGCATCTT3′; and IL-1, 5′-GCTGAAGGAGTTGCCAGAAA-3′ and 5′-GTGCAAGTGACTCAGGGTGA-3′. One PCR cycle consisted of the following: 94°C for 30 s, 60°C for 30 s, and 72°C for 30 s. After standardization to the housekeeping gene β-actin, relative expression of each gene compared with that in control animals was calculated according to the manufacturer’s instructions.

Serum was deproteinized with an equal volume of 6% 5-sulfosalicylic acid. Concentrations of l-Arg and total reactive nitrogen metabolites were determined in the serum, by HPLC analysis and chemiluminescence with an NO analyzer (NOA 280; Sievers, Boulder, CO), respectively, as previously described (18).

Data are expressed as the mean ± SEM. When comparisons between multiple groups were made, the Student-Newman-Keuls test was used, and for comparisons between two groups, Student’s t test was used. For survival data, χ² or Fisher’s exact test was used as appropriate at each time point, and Cox hazard regression analysis was used to determine the effect of l-Arg throughout. SAS (SAS Institute, Cary, NC) was used for the Cox regression, and Statview v. 5.01 for the Macintosh (SAS Institute) was used for all other analyses.

As shown in Fig. 1,A, expression of both arginase I and arginase II was very low in control tissues. Arginase I was increased in the colon of C. rodentium-infected WT mice, while arginase II mRNA was not induced. By real-time PCR, arginase I mRNA levels of infected mice were increased by 9.7 ± 2.5-fold compared with uninfected mice (n = 6 for each; p < 0.01), while there was no significant change in arginase II mRNA levels from the low control levels (data not shown). In addition, iNOS mRNA levels were also found by RT-PCR (Fig. 1,A) and by real-time PCR (data not shown) to be markedly increased in the colon of mice infected with C. rodentium, and absent in the colon of uninfected mice. As shown in Fig. 1 B, arginase I protein was present in the tissues of infected mice and absent from control, while arginase II levels were very low in both control and infected mice.

Focally intense staining for arginase I in colonic tissues of infected mice is shown in Fig. 2, B, C, and H. Staining was present in epithelial cells (Fig. 2, B and C), but was also found throughout the mucosa in severe colitis (Fig. 2,H), with staining of infiltrating inflammatory cells. In serial sections from the same tissues, iNOS localized to the epithelium, and the lamina propria and submucosal inflammatory cells (Fig. 2, E, F, and I). Staining was absent in uninfected tissues with Ab to arginase I (Fig. 2,A) or iNOS (Fig. 2,D) or in infected tissues incubated with an isotype Ig control (Fig. 2 G). iNOS staining was completely absent in tissues from iNOS^−/− mice (data not shown).

In the colon of infected animals, arginase activity was increased by ∼2.4-fold vs control mice (Fig. 3,A). A significant increase in NO concentration was observed in the serum of infected mice compared with control mice (Fig. 3,B). A concomitant marked decrease of l-Arg concentration was observed in the serum of infected mice (Fig. 3 C).

Because arginase I and iNOS were both abundantly expressed in the colon of C. rodentium-infected mice, and l-Arg was completely metabolized, we investigated the effect of l-Arg supplementation. In WT mice, C. rodentium infection was associated with a high level of mortality that began on day 9 postinfection (Fig. 4,A). When WT animals were treated with l-Arg, mortality was inhibited by 42 and 62% compared with mice receiving water alone, after 12 and 14 days of infection, respectively (Fig. 4,A). To further assess the impact of l-Arg on survival, Cox regression analysis was performed. WT mice treated with l-Arg had only a 31% hazard of death compared with mice receiving water alone (p < 0.0009). In iNOS^−/− mice, no deaths were observed, with or without l-Arg (Fig. 4 A).

In WT mice, l-Arg treatment reduced weight loss, and iNOS-deficient mice given l-Arg had further improvement, actually gaining weight in the presence of infection (Fig. 4,B). The weight loss of WT mice is actually underestimated, because only the weights of animals still alive could be included. Colon weight was significantly increased by >2-fold in C. rodentium-infected WT mice (Fig. 4 C). Colonic weight was decreased by 28% in the WT l-Arg group, and by 38% in the iNOS^−/− mice. An additive effect of iNOS deletion and l-Arg administration was observed, with a 68% decrease in colon weight. Neither iNOS deletion nor l-Arg treatment had an effect on the colon weight of uninfected control mice.

When compared with control mice (Fig. 5,A), WT mice infected with C. rodentium (Fig. 5,B) developed changes that included transmural inflammation, ulceration, mucosal bacterial aggregates, and epithelial hyperplasia with mucin depletion. There was reduction of colitis in iNOS-deficient mice (Fig. 5,C), with less colonic thickening and inflammatory cell infiltration, and l-Arg further ameliorated the inflammation, epithelial reactivity, and mucin depletion (Fig. 5,D). Because individual scores for acute inflammation, chronic inflammation, and epithelial reactivity had similar patterns, a composite score is shown in Fig. 5 E. iNOS deletion and iNOS deletion + l-Arg inhibited the mean histologic injury score by 48 and 72%, respectively, compared with WT mice. l-Arg treatment of iNOS^−/− mice resulted in a 46% improvement compared with water-treated iNOS^−/− mice. It should be noted that the mean histologic injury score in the WT mice is likely to be underestimated, due to the early mortality that precluded obtaining histologic samples in some of the mice that would be expected to have had high injury scores. In uninfected mice, no differences were observed between the colons of WT and iNOS^−/− mice, with or without l-Arg treatment (data not shown).

Consistent with the decrease in histopathology, there was a significant reduction in C. rodentium colonization levels in the iNOS^−/− mice (WT mice, 8.3 ± 3.0 × 10⁹ CFU/g colon, n = 12; iNOS^−/− mice, 5.6 ± 3.3 × 10⁶ CFU/g colon, n = 18; p < 0.01 vs WT). However, l-Arg supplementation did not alter colonization levels in WT mice (WT l-Arg, 3.2 ± 1.7 × 10¹⁰ CFU/g colon, n = 12), and actually increased levels in iNOS^−/− mice compared with mutant mice receiving water alone (iNOS^−/− l-Arg, 2.8 ± 1.8 × 10⁸ CFU/g colon, n = 13; p < 0.05 vs iNOS^−/−), indicating that the clinical and histologic improvement with l-Arg is not simply due to inhibition of bacterial growth by this amino acid.

Because C. rodentium colitis has been strongly associated with activation of the Th1 cytokine IFN-γ, and the associated proinflammatory cytokines TNF-α and IL-1 (19, 20), we sought to determine the relationship between the clinical and histologic effects that we observed with these immunologic parameters. Real-time PCR analysis demonstrated a marked increase in IFN-γ (Fig. 6,A), TNF-α (Fig. 6,B), and IL-1 (Fig. 6 C) in C. rodentium colitis tissues compared with normal tissues, and a significant attenuation of these increases with either iNOS deletion or l-Arg treatment.

Because ornithine, the product of arginase, is metabolized by ODC to form polyamines, we investigated ODC expression in the colon of C. rodentium-infected WT mice. By real-time PCR, we found a 2.5 ± 0.5-fold increase of mRNA level in infected mice (n = 13) compared with control mice (n = 4; data not shown). However, a ∼40-fold increase of ODC activity was measured in the colon of either infected WT or WT l-Arg mice (Fig. 7,A). This increase was not likely to be due to ODC activity from C. rodentium itself, because we measured bacterial ODC activity, and determined that it represented no more than 1% of the total ODC activity in the tissue. There was a 1.8 ± 0.1-fold increase in colonic polyamines in C. rodentium-infected WT mice, and a significant, further increase of 2.7 ± 0.2-fold with l-Arg treatment (Fig. 7 B), indicating that arginase activity was an important determinant of polyamine synthesis.

To further demonstrate the beneficial effect of arginase and polyamine formation, we conducted experiments with BEC and DFMO, inhibitors of arginase and ODC, respectively. In uninfected control mice, BEC or DFMO treatment had no effect (Table I). However, there was a significant loss of survival in C. rodentium-infected WT mice treated with BEC or with DFMO (Table I). In fact, the experiments needed to be terminated at 10 days after infection, because of the deaths and severe disease at this point. The colons of C. rodentium-BEC and C. rodentium-DFMO groups had a greater increase in weight and histologic injury than those of the C. rodentium-water group (Table I). When compared with the infected WT mice treated with water (Fig. 8,A), the colons of both BEC (Fig. 8,B)- and DFMO (Fig. 8 C)-treated mice showed marked transmural inflammation and mucin depletion. The BEC-treated mice had substantial submucosal abscess formation, and the DFMO-treated mice exhibited mucosal and submucosal hemorrhage, both indicative of severe acute inflammation.

To further confirm the importance of arginase, BEC was administered to iNOS^−/− C. rodentium-infected mice. When compared with iNOS^−/− alone, BEC caused a significant worsening of colon histologic injury scores (iNOS^−/−: 2.42 ± 0.46, n = 12 vs iNOS^−/− + BEC: 6.44 ± 0.98, n = 8, p < 0.01) and colon weight (iNOS^−/−: 0.36 ± 0.01% of total body weight vs iNOS^−/− + BEC: 0.92 ± 0.20%, p < 0.05).

Our data demonstrate for the first time an important effect of arginase in the intestine in vivo, namely a beneficial role of the arginase-ODC metabolic pathway in colitis. Arginase I and iNOS are two enzymes up-regulated in C. rodentium colitis that metabolize l-Arg. Because iNOS deletion ameliorated colitis, and l-Arg supplementation was protective in both WT and iNOS^−/− mice, our data indicate that arginase exerts its salutary effects by increasing polyamines in addition to inhibiting NO production. This was confirmed by the in vivo use of BEC, an arginase inhibitor. The improvement in animals fed l-Arg is most likely due to enhanced formation of polyamines, because an increase of polyamine concentration was demonstrated in l-Arg-treated mice, and ODC inhibition with DFMO worsened the disease.

The fact that l-Arg treatment resulted in improvement most likely derives from the systemic depletion of l-Arg that occurred in the C. rodentium-infected mice. We measured serum l-Arg rather than tissue levels, because l-Arg is in constant flux between the intracellular and extracellular space (25), making serum levels a more reliable means to measure l-Arg availability. As an example of this, in mice with targeted deletion of arginase II, the metabolic effect was measured in the serum in which significant accumulation of l-Arg was demonstrated in these mice (26).

It should be noted that there was a marked (40-fold) increase in ODC activity in C. rodentium colitis tissues, but only a 2-fold increase in polyamine levels. This is due in part to the fact that polyamines are rapidly acetylated, which results in their efflux out of cells and excretion (27). Additionally, polyamine synthesis is dependent on availability of l-ornithine substrate for ODC, which derives from l-Arg by the activity of arginase; our current data show that arginase activity is increased only ∼2-fold in the colitis tissues, so this can be rate limiting in polyamine synthesis. Another important point is that DFMO is an irreversible inhibitor of ODC (28), and ODC is absolutely required for polyamine synthesis. Therefore, at an effective dose of DFMO, supplementation of l-Arg or iNOS deletion would not be expected to reverse the effect of DFMO on polyamine synthesis or the exacerbation of colitis.

The potent inhibition of ODC by DFMO is also illustrated by the more severe exacerbation of colitis by DFMO than by BEC. It is likely that C. rodentium infection has a direct effect on ODC induction. Because we observed a 2.5-fold increase in ODC mRNA, there is activation of mRNA expression, but because of the 40-fold increase in enzyme activity, it is likely that there may be posttranslational effects on ODC activity as well. Direct activation of ODC by C. rodentium is also supported by our findings of an 87-fold increase in ODC activity in intestinal epithelial-6 cells stimulated in vitro with C. rodentium for 4 h (our unpublished data).

To our knowledge, this is also the first report of the activation of arginase I in the gastrointestinal tract under pathophysiologic conditions. We have reported that arginase II mRNA and protein levels were up-regulated in the stomach of Helicobacter pylori-infected mice and humans, but arginase I was not induced in these tissues (11). Murine macrophages and dendritic cells express arginase I when stimulated by Th2 cytokines (29), LPS (30), or cAMP (15). Even though C. rodentium infection is a Th1-driven disease, with no increase in Th2 cytokines (20) (our own results), arginase I expression was induced in this model. We have also found that in vitro, arginase I is induced in C. rodentium-stimulated murine macrophages (our unpublished data), consistent with expression in lamina propria cells in our immunohistochemistry data. It is possible that the selective expression of arginase I may be due to: 1) a specific activation by C. rodentium and/or its released factors; 2) a colon-specific expression; or 3) selective induction in an acute form of inflammation vs chronic inflammation. In studies in mouse DSS (31) and rat TNBS colitis tissues (our unpublished data), we also found that arginase I, and not arginase II, is up-regulated; therefore, the expression of arginase I in colitis is not specific to C. rodentium. Because it has been shown that arginase I is up-regulated in a mouse model of acute immune complex-induced inflammation of the nephron (32) and in macrophages from an acute wounded rat model (30), the expression of arginase I is not specific to the colon. It is also unlikely that arginase I occurs only in acute inflammation, because we have observed increased expression of arginase I in human inflammatory bowel disease tissues (discussed below), in which there is a chronic component to the disease and the histologic injury pattern.

It is likely that the main effect of l-Arg treatment is enhanced substrate availability for both arginase and iNOS. It should be noted that we demonstrated increased arginase activity in the colitis tissues, while showing increased iNOS activity by measuring serum NO production. There is abundant published evidence that iNOS enzymatic activity is increased in rodent models of colitis and in human inflammatory bowel disease and that this correlates with iNOS expression; iNOS activity is regulated in vitro and in vivo by enzyme abundance (see Ref. 33 for our review on this subject). We measured serum NO production to yield a more sensitive means to determine the relationship of l-Arg consumption to iNOS activity. The fact that the histologic scores of infected WT l-Arg-treated mice are not significantly different from those of water-treated mice could be due to an overproduction of NO, counteracting the beneficial effect of arginase metabolism and polyamine synthesis. Because NO can inhibit ODC by S-nitrosylation (34), this could interfere with colonic ODC activity.

Polyamines have been shown to have numerous biological functions that may be relevant to amelioration of colitis, including inhibition of monocyte activation and Th1 cytokine production (35), stimulation of epithelial restitution by enhancing cell migration (36) and proliferation (37), and regulation of apoptosis (38). It has been demonstrated that polyamine production is essential for the repair of rat duodenal mucosa after stress (39), and that ODC activity is increased in the acetic acid colitis model (40). Arginase activity can also result in increased collagen production through proline synthesis from l-ornithine by the enzyme ornithine aminotransferase (25). Collagen production and polyamine synthesis have both been proposed as major events leading to wound healing by favoring fibroblast replication (41).

We have addressed the potential effect of modulation of polyamine levels on immune response by measuring Th1 and proinflammatory cytokines and demonstrating that the large increases in IFN-γ, TNF-α, and IL-1 in C. rodentium colitis are ameliorated by either l-Arg treatment or iNOS deletion. These findings may be attributable to enhanced substrate availability for polyamine synthesis in these groups; however, it should be recognized that these cytokines are markers of inflammation and our results may simply represent a correlation with the improvement of colitis. We have also found that there is induction of epithelial apoptosis in C. rodentium colitis tissues by TUNEL assay (data not shown) and a significant increase in apoptosis in vitro in intestinal epithelial-6 cells stimulated with C. rodentium (our unpublished data). The interaction of epithelial apoptosis, proliferation, and cell migration in response to C. rodentium with acute vs chronic infection is likely to be important and is a complex area that requires additional investigation.

Similar to our findings, iNOS expression was shown to be up-regulated in epithelial and lamina propria cells of mice infected with C. rodentium in a recent report (42). However, there was no reduction in bacterial colonization or of colonic weight or histologic injury in iNOS^−/− mice in this study, although there was an improvement in survival that was not statistically significant (42). We attribute these differences from our results in the current study of significant improvement of each of the above parameters in iNOS^−/− mice to several possibilities, including their use of a lower inoculum of C. rodentium, younger age of infected mice (3–4 wk vs 8 wk in our study), and specific pathogen-free housing vs conventional housing in our study. A second report showed no differences between WT and iNOS^−/− mice (43), but in that study a different strain was used (E. coli attaching and effacing (eae)-deficient mutant complemented with eae from EPEC) and colonization studies revealed >2 log orders less C. rodentium/g of tissue than we recovered at day 14 in WT mice, indicating important differences in their model. Our data show a correlation of decreased colonization levels of iNOS^−/− mice with decreased histologic injury, proinflammatory cytokine levels, and clinical course. However, there is a more complex relationship between colonization and indicators of disease, because l-Arg led to improvement of both WT and iNOS^−/− mice without decreasing colonization. Our present results are strengthened by the fact that we have obtained similar preliminary results in the DSS model of colitis (31). l-Arg improved, while BEC worsened clinical and histologic features of colitis in both WT and iNOS^−/− mice (31), indicating that the findings of a beneficial role of arginase are not limited to one model of colitis.

In human inflammatory bowel disease, an increase in ODC activity has been described in ulcerative colitis and Crohn’s disease in children (44), while a decrease in ODC activity has been reported in severe human colitis (45). Interestingly, a decrease in ODC activity, but an increase in mucosal spermidine concentration has been reported in patients with ulcerative colitis (46). We have found that expression of arginase I, arginase II, and ODC mRNA is up-regulated in both human ulcerative colitis and Crohn’s disease tissues (our unpublished data), providing relevance for the murine colitis studies presented in this work. Additionally, induction of arginase II by cAMP has been described in human colonic epithelial cells (47); we have observed that cAMP can also induce arginase I expression in colonic cell lines (our unpublished data).

In summary, our data define an important in vivo role for arginase as a regulator of mucosal inflammation. In addition to synthesis of polyamines in the colonic mucosa, alterations in luminal uptake, as well as polyamine metabolism and efflux may be important events in colitis. Chronic overproduction of polyamines could contribute to the risk for colorectal cancer associated with colitis due to the growth-promoting effects of polyamines and the oxidative stress that occurs with polyamine metabolism by polyamine oxidase. Nonetheless, enhancement of arginase and ODC activities may represent important new strategies for ameliorating inflammatory bowel disease.