Psychological stress reactivates dextran sulfate sodium-induced chronic colitis in mice

Abstract

Inflammatory bowel disease (IBD) is a chronic condition with alternating active and quiescent phases of inflammation. Stress has been suggested as a factor triggering a relapse of IBD. We investigated the role of repetitive psychological stress [water avoidance stress (WAS)] in reactivating colonic inflammation in a murine model of dextran sulfate sodium (DSS)-induced chronic colitis. Colitis was induced in C57BL/6 female mice by exposure to 3% DSS (5 days). During chronic inflammation (day 34), mice underwent repetitive WAS (1 h/day/7 days) and were given a sub-threshold concentration of DSS (1%, 5 days) or normal water to drink. At euthanasia (day 40), inflammatory parameters were assessed (colon inflammatory score, levels of inflammatory markers and histology). Mice with chronic colitis exposed to WAS had higher macroscopic and microscopic colonic inflammatory scores and levels of inflammatory markers (mainly IL-1β, IL12p40 and CCL5) than non-stressed mice. Inflammatory responses were further enhanced by the presence of a sub-threshold concentration of DSS (1%). In mice without chronic inflammation, neither WAS nor 1% DSS, individually or in combination, elicited any inflammation. Hence stress, per se, reactivates a quiescent chronic inflammation, but does not initiate inflammation in healthy mice. Stress should be regarded as an environmental factor triggering IBD relapses in humans.

• Chemokines
• colitis
• cytokines
• inflammation
• inflammatory bowel disease
• water avoidance stress

Introduction

Inflammatory bowel diseases (IBD), including Crohn's disease (CD) and ulcerative colitis (UC) are chronic gastrointestinal inflammatory conditions characterized by alternating phases of remission and recurrence of symptoms, without a defined pattern and with no explained cause for the relapses (Podolsky 2002; Baumgart and Carding 2007; Xavier and Podolsky 2007). Although the aetiology remains unknown, the pathogenesis is likely dependent on the interaction between local immune reactions within the intestinal wall and environmental factors in genetically susceptible individuals (Andus and Gross 2000; Baumgart and Carding 2007; Xavier and Podolsky 2007).

Psychological stress is an environmental factor that has long been suggested to contribute to the pathophysiology of IBD. Indeed, during the 1950s, UC was regarded as a psychosomatic disease (Engel 1969). While clinical observations have provided anecdotal evidence (Salem and Shubair 1967; Robertson et al. 1989), few prospective studies have examined whether or not stress is involved in the exacerbation or precipitation of inflammatory relapses in IBD patients. Nevertheless, recent studies suggest that several stressors can increase the rate of relapse in patients with IBD (Mawdsley et al. 2006). A few studies in animal models of colitis also suggest that stress interacts with the course of inflammation and, specifically, aggravates inflammatory responses, as determined by the assessment of several inflammatory markers [such as the content of pro-inflammatory chemo- and cytokines or the expression of nitric oxide synthase (NOS) and cyclooxygenase-2 (COX-2)] (Collins et al. 1996; Gue et al. 1997; Qui et al. 1999; Million et al. 1999; Reber et al. 2006; Saunders et al. 2006; Ponferrada et al. 2007). However, these studies have concentrated on the stress–inflammation interaction during the acute phase of inflammation. So far, no study has addressed the role of stress in modulating the course of chronic intestinal inflammation and, in particular, the reactivation of chronic inflammation; a condition that will reflect more accurately the relapse of inflammation observed in IBD patients.

The dextran sulfate sodium (DSS)-induced colitis model is a well-established animal model with clinical and histopathological features in the acute phase resembling UC (Okayasu et al. 1990). Moreover, we have reported that a single exposure to DSS for 5 days induces a chronic inflammation in C57BL/6 mice, observed up to 4 weeks after DSS removal (Melgar et al. 2005, 2006). This state of chronic colitis was characterized by persistent histopathological changes (infiltrate of mononuclear cells, irregular epithelial structure and persistent deposits of collagen) and a moderate, but sustained, production of colonic inflammatory mediators [such as IL-1, IL-12p40, RANTES/CCS and interferon-γ (IFN-γ] (Melgar et al. 2005, 2006).

In this study, we assessed the role of repetitive psychological stress water avoidance stress (WAS), in reactivating colonic inflammation in a murine model of DSS-induced chronic colitis (Melgar et al. 2005). Since WAS has not been used in a systematic manner in mice, we also assessed the validity of such a stimulus as a suitable psychological stressor in this species. Moreover, we also evaluated if a sub-threshold inflammatory stimulus is necessary for stress to have an effect or if stress, per se, is able to reactivate a chronic inflammatory state. The responses to stress were determined by assessing faecal pellet output, acute changes in body weight and plasma corticosterone levels and adrenal weight. The assessment of inflammatory responses was based on clinical symptoms, plasma and colonic inflammatory markers and histology. The pro-inflammatory chemo- and cytokines analysed [e.g. IL-1, IL-6, chemokine (C-X-C motif) ligand 1 (CXCL1)] have been shown to play a role in acute inflammatory responses in the experimental models of IBD as well as in IBD patients (Podolsky 2002; Baumgart and Carding 2007; Xavier and Podolsky 2007; Melgar et al. 2005, 2006; Niess et al. 2002).

Materials and methods

Animals

Adult C57BL/6JOlaHsD female mice (20–25 g; Harlan, Horst, The Netherlands) were used. Mice were group-housed in standard cages (six mice per cage) and maintained in a 12 h light: 12 h darkness cycle (lights on at 06:00 h) and under controlled conditions of temperature (21°C) and humidity (50%), with ad libitum access to a standard diet (R3 pellets, Lactamin, Stockholm, Sweden) and tap water, except when receiving DSS. Mice were allowed to acclimatize to the animal facility for at least 2 weeks before starting the studies. All protocols were approved by the local animal ethics review committee in Gothenburg (Gothenburg University), Sweden.

Induction of colitis

A solution of DSS (45 kDa; 1 or 3% concentration in water; TdB Consultancy AB, Uppsala, Sweden) was used to induce colitis. Fresh DSS solutions were prepared daily during the 5-day treatment period. Similar protocols have been used in previous studies in mice to induce colitis (Melgar et al. 2005, 2006; Reber et al. 2006). Normal tap water was used as the control treatment.

Psychological stress (water avoidance stress, WAS)

Water avoidance was used as a psychological stressor and was performed following procedures described for rats, with minor modifications (Martinez et al. 1997; Cameron and Perdue 2005; Yang et al. 2006). Briefly, the mice were placed on a platform (4 cm diameter, 6 cm height) located in the centre of a bucket (35 cm diameter, 54 cm height) filled with tap water (17.5–18.5°C) to about 1 cm below the edge of the platform. To avoid contact with the water, mice stand on the platform for the duration of the experimental period. Stress sessions lasted 1 h and were repeated on 7 consecutive days (1 h/day). Mice maintained in non-stressful conditions (sham stress) served as controls. Sham-stressed mice were transferred as a group to a clean cage, otherwise identical to their home cage, maintained in their normal housing environment for 1 h and thereafter returned to their original home cage. All procedures were performed in the morning (ending no later than 12:00 h) to minimize any influence of circadian rhythms. Faecal pellet output and acute changes in body weight during the 1 h session of WAS/sham stress were used as markers of stress. In all cases, body weight was measured before and after each WAS/sham session. Pellet output was assessed individually for each animal exposed to WAS, while counted on a group basis for the sham stress groups.

Experimental protocols

Mice (n = 48) were randomly divided into eight experimental groups (n = 6 per group, Table I). For every experimental group, all mice were housed together at all times. In a random assignment, the experimental groups received tap water or a solution of 3% DSS during a 5-day period (days 0–5, Figure 1) followed by a 29-day recovery period, during which all mice received normal water. Between days 34 and 40 mice were exposed to a daily session of WAS or sham stress (1 h/day). In addition, starting on the second day of WAS/sham stress (day 35), some groups received a solution of 1% DSS until the moment of euthanasia, on day 40, immediately after the last session of WAS/sham stress (Figure 1). Individual body weight, the general state and the presence of clinical signs of inflammation were assessed on a daily basis throughout the study.

Table I.  Experimental groups.

Experimental group	Treatment 1 (days 0–5)	Treatment 2 (days 34–40)	Treatment 3 (days 35–40)
1	3% DSS^*	WAS^†	1% DSS
2	Water	WAS	1% DSS
3	Water	Sham stress	Water
4	3% DSS	Sham stress	1% DSS
5	Water	Sham stress	1% DSS
6	Water	WAS	Water
7	3% DSS	Sham stress	Water
8	3% DSS	WAS	Water

^* Dextran sulfate sodium

^† Water avoidance stress.

Figure 1 Overview of the experimental protocols used in the studies. Mice received 3% DSS or tap water during an initial period of 5 days (0–5). Thereafter, they were given time to develop a mild chronic inflammatory condition (days 6–34) before starting repetitive exposure to stress (WAS) or sham stress (Sham) for a 7-day period. In addition, during the WAS/sham stress period, some mice were exposed to a 1% DSS solution for the final 5 days (days 35–40). On day 40, all mice were euthanized and inflammatory parameters assessed.

Tissue and plasma sampling

At the time of euthanasia, mice were deeply anaesthetized with isoflurane (Abbot Scandinavia, Solna, Sweden) and blood samples were taken by cardiac puncture. Blood samples were collected in EDTA-coated tubes, centrifuged and the plasma collected and stored at − 80°C until analysis. After blood sampling, the abdominal cavity was opened and the colon carefully dissected, cut close to the ileocecal junction and the rectum, and de-inflammation assessed macroscopically of the colon. Thereafter, a 3 cm segment of the most distal part of the colon was cut off, weighed and divided into two longitudinal sections. One of the sections was fixed in zinc–formalin solution (pH 7.4; Histolab Products, Gothenburg, Sweden) for histological studies. The other section was weighed, frozen in liquid nitrogen and stored at − 80°C until analysis. In all cases, the adrenal glands, the thymus and the spleen were dissected free and weighed.

Clinical and macroscopic assessment of inflammation

Clinical assessment of inflammation included daily monitoring of body weight, appearance of faeces and general health condition. Separate scores (0–2) were assigned for faecal consistency and health condition (including hunch posture and piloerection); where 0 indicates normal faecal content/healthy condition, 1 indicates loose faecal content/signs of hunch posture and/or piloerection and 2 indicates watery diarrhoea/severe hunch posture and piloerection. At necropsy, the macroscopic appearance of the colon was scored, in a blind manner, based on the presence of inflammatory signs (inflammatory score), consistency of faecal content (score 0–3) and presence of visible faecal blood (score 0–3) as previously published (Melgar et al. 2005). The colonic inflammatory score was based on the extent of oedema (0–3), thickness (0–4), stiffness (0–2) and presence of ulcerations (0–1), resulting in a maximum total score of 10.

Histological studies

Paraffin sections (5 μm) of tissue samples were stained with hematoxylin-eosin following standard histological procedures. Coded sections were observed with a light microscope (Axioskop, Zeiss, Germany) in a blinded fashion and the epithelial structure and the presence of ulcerations and the inflammatory infiltrate were assessed. Based on these parameters a histopathological score (0–5) was assigned to each sample: 0, normal tissue; 1, mild alterations; 2, mild-to-moderate alterations; 3, moderate alterations; 4, moderate-to-severe alterations; 5, severe alterations. In addition, morphometric characteristics (thickness of the mucosa–submucosa region and muscle layers) were assessed in 10 randomly selected areas, covering the whole tissue thickness (10 × magnification), of each section using image analysis software (Image Access Analysis, EuroMed Networks, Stockholm, Sweden).

Analysis of stress, inflammatory and metabolic markers in plasma and colonic homogenates

Local inflammatory markers [interleukin 1β (IL-1β); IL-6; IL-12p40; tumour necrosis factor-α (TNF-α); chemokine C-X-C motif ligand 1 (CXCL1); cytokine regulated upon activation, normal T-cell expressed, and presumably secreted-RANTES/CCL5 (CCL5) and JE/mouse monocyte-chemoattractant protein-1-MCP-1/CCL2 (CCL2)] were determined in whole colonic tissue homogenates. Tissue samples were homogenized in buffer containing phosphate sodium saline (PBS; Gibco, Invitrogen, UK), 10% fetal calfserum (Gibco) and protease inhibitor cocktail tablets (Complete Mini, Roche Diagnostics GmbH, Mannheim, Germany), as described recently (Melgar et al. 2005, 2006). Commercial ELISA kits (R&D Systems Europe, Abingdon, UK) were used to determine the concentrations of IL-1β (code MLB00B; limit of detection, 3 pg/ml; intra-assay variability, 1.5–4.4%; inter-assay variability, 2.8–6.1%) and IL-6 (code R600B; limit of detection, 21 pg/ml; intra-assay variability, 4.5–8%; inter-assay variability, 7–10%), according to the manufacturer's instructions. Levels of IL-12p40 TNF-α, CXCL1, CCL5 and CCL2 were determined using the xMAP^® technology developed by Luminex^® (Luminex Corporation, Austin, TX, USA) as described previously (Melgar et al. 2005). The range of detection in the xMAP^® technology is 1–32,000 pg/ml with inter- and intra-assay variabilities < 10%. In all cases, protein levels are expressed as picograms per 100 mg colonic tissue.

Plasma concentrations of the acute-phase protein haptoglobin were determined using a commercial colorimetric assay kit for haptoglobin (TP 801, limit of detection, 0.02 mg/ml; intra-assay variability, 0.88–1.30%; inter-assay variability, 1.68–4.6%; Tridelta Development LTD, Ireland). Plasma concentrations of corticosterone, as a marker of stress, were determined by radioimmunoassay using a commercial kit (RPA 548, limit of detection, 0.06 ng/tube; intra-assay variability, 5%; inter-assay variability, 4–5.9% Biotrak Assay Systems, Amersham, UK).

Plasma metabolic parameters (triglycerides, cholesterol and glucose) were determined using the following commercial kits: Roche Diagnostics nos. 12146029216 (triglycerides) and 2016630 (cholesterol) (Mannheim, Germany) and ABX Diagnostics-Parch Euromedicine no. HK 125 for glucose (Montpellier, France).

mRNA expression of inflammatory markers

The expression levels of mRNAs for IL-6, IFN-γ, TNF-α, COX-2 and the inducible form of the nitric oxide synthase (iNOS) were assessed on whole colonic tissue homogenates using real-time PCR (Taqman^®). Tissue samples were homogenized with Mixer Mill (Retsch GmbH, Haan, Germany) in Trizol (Invitrogen AB, Lidingö, Sweden) followed by total RNA isolation, according to the manufacturers' instructions. For each cDNA synthesis, 500 ng DNAase-treated total RNA was used with the High-Capacity cDNA archive kit (Applied Biosystems). Each sample was run in triplicate with 4 ng template in each reaction (10 μl). Data were normalized by using the internal standard, 36B4 (acidic ribosomal phosphoprotein P0), and expressed as “relative expression of the current gene”. Primers and probes for the internal control were designed from cDNA sequences in Primer ExpressTM 2.0 (Applied Biosystems) and purchased from Operon (Cologne, Germany) (Probe: 5′-TCG GTC TCT TCG ACT AAT CCC GCC AA-3′). The assays for the genes of interest were purchased as TaqMan^® Gene Expression Assays (Applied Biosystems) (TNF-α, Mm00443258_m1; COX-2, Mm00478374_m1; IFN-γ, Mm00801778_m1; IL-6, Mm00446190_m1; iNOS, Mm00440485_m1).

Statistical analysis

Results are expressed as group mean ± SEM. Comparisons between multiple groups were performed using one-way analysis of variance (ANOVA) or a Kruskal–Wallis one-way ANOVA on ranks, as appropriate, followed, whenever necessary, by a Student–Newman–Keuls multiple comparisons test. Comparisons between two groups were performed using a Student's t-test. P-values < 0.05 were considered statistically significant. In addition, to assess the relationship between all the variables tested (31 variables in 48 mice) a multivariate analysis, based in he principal components analysis (PCA), as implemented in the SIMCA-P software v. 11.0 (SIMCA at Umetrics, Umeå, Sweden), was also performed (Jackson 1991).

Results

Clinical indices of inflammation during the induction of chronic colitis

For the determination of clinical indices of inflammation during the induction of chronic colitis (days 0–34), all the groups receiving normal water or 3% DSS were merged into two experimental groups (n = 24 per group). Mice receiving normal water throughout the experiment period showed no clinical signs of inflammation or spontaneous intestinal inflammation. Mice receiving normal water during days 0–5 showed a steady and linear increase in body weight throughout the experimental period (Figure 2A). However, from day 5, mice exposed to 3% DSS showed a significant loss of body weight when compared with their weight at the beginning of the study or the weight in control mice receiving normal water (Figure 2A). Peak body weight loss was observed on day 8 (9.8 ± 1.4% loss; P < 0.05 vs. initial body weight). Thereafter, the body weight recovered slowly, at day 17 reached values similar to those at day 0 and by day 34, the mean body weight was similar in controls (24.1 ± 0.2 g) and 3% DSS-exposed mice (23.6 ± 0.3 g; P>0.05) (Figure 2A). The daily assessment of the general health condition and faecal consistency followed time profiles similar to those described for the body weight changes, with peak scores reached between day 7 and 10 post-DSS exposure (Figure 2B, C). Thereafter, the general health condition was completely normalized by day 17, while the faecal score was maintained slightly increased over the control group through the experimental period (Figure 2B, C). These clinical observations are consistent with the development of acute colitis between days 6 and 10 after the exposure to 3% DSS and the progression to chronicity, as previously described by us (Melgar et al. 2005).

Figure 2 Effects of exposure to 3% DSS on clinical symptoms: (A) body weight, (B) general health condition and (C) faecal consistency were assessed as clinical indicators. Data are group mean ± SEM and correspond to pooled data from all mice receiving 3% DSS for 5 days (n = 24) or normal water (n = 24) throughout the experimental days 0–33. The shadowed area represents the 5-day period of 3% DSS exposure.

Effects of repetitive psychological stress on colonic motor activity and body weight

When compared with the sham stress groups, all experimental groups subjected to repeated WAS showed a significant increase in their rate of defecation during the duration of stress (Figure 3A, B). Responses to WAS were stable over the 7-day duration of stress. In mice receiving normal water, the mean defecatory frequency during stress was increased by 12-fold over the rate in sham stress controls (P < 0.05). This response was attenuated when mice were exposed to 1% DSS in combination with stress, and further reduced in mice previously exposed to 3% DSS, independently of the treatment applied in combination with the stress (i.e. normal water or 1% DSS) (Figure 3A, B). In mice maintained in non-stressful conditions, regardless of the treatments applied, pellet output was low (Figure 3A, B).

Figure 3 Effects of repeated water avoidance stress (WAS, 1 h/day for 7 days) on faecal pellet output and body weight. (A) Mean faecal pellet output during the time of stress across the 7 days of treatment. *P < 0.05 vs. respective sham stress group. +P < 0.05 vs. the water + water group. (B) Cumulative pellet output for the 7-day period of stress. *P < 0.05 vs. respective sham stress group. +P < 0.05 vs. other WAS groups (ANOVA). (C) Acute change in body weight during the 1 h WAS for the 7 days of stress (% change from pre-stress weight). In B and C, the dotted lines with open symbols correspond to the responses in the respective sham stress groups.

During WAS, regardless of previous exposure to 3% DSS or water, there was a significant acute body weight loss, which was not detected in the sham stress groups, except for the first day of the procedure (Figure 3C). Weight loss in the groups exposed to chronic stress was only transient as the mice recovered, and in most of the cases gained weight at the time of the next WAS session. No correlation between the number of faecal pellets excreted during the time of stress and the body weight changes was observed. Since the weight of the faecal pellets excreted could not be assessed in the stress groups due to the experimental conditions, the faecal mass excreted was estimated by weighing 20 groups of fresh faecal pellets (10 faecal pellets included in each group) randomly collected from the normal home cages at the time of the experiments (mean weight of 10 faecal pellets: 226.6 ± 6.8 mg). This amount of faeces could account for about 50% of the total body weight loss observed during stress exposure.

Overall, body weight during the 7-day treatment period (days 34–40) was relatively stable. Nevertheless, net increases in body weight were, in general, lower in the stressed mice when compared with their respective sham-stressed counterparts (data not shown).

Clinical symptoms during WAS and at the time of necropsy

The daily general health condition, body weight or faecal consistency of healthy or chronically inflamed mice was not significantly affected by repetitive WAS. Nevertheless, mice subjected to WAS had higher faecal scores than their respective sham stress controls (Figure 4A). This was clear for the water–WAS–1% DSS and the 3% DSS–WAS–water groups (both P < 0.05 vs. respective sham stress group), while only a tendency was observed for the 3% DSS–WAS–1% DSS mice (P = 0.087) (perhaps because in the respective sham stress group, faecal score was already increased when compared with all other sham stress groups) (Figure 4A). No signs of faecal blood were observed in the colon of either mice exposed to 3% DSS or normal mice with or without WAS.

Figure 4 Colonic inflammatory and clinical scores at the time of necropsy in the different experimental groups. (A) Faecal consistency. (B) Macroscopic inflammatory score. (C) Colon length. (D) Relative colon weight. Data are mean ± SEM values for six mice per group. *P < 0.05 vs. respective sham stress group. +P < 0.05 vs. other sham stress groups (ANOA) (ANOVA).

Plasma concentrations of corticosterone and metabolic markers

In mice receiving normal water, exposure to stress increased the plasma concentrations of corticosterone by threefold over the levels in non-stressed mice (P < 0.05; Table II), as determined immediately after the last stress session. Corticosterone levels in sham-stressed groups receiving DSS were increased over the values in sham-stressed non-inflamed (water-exposed) mice in a graded manner. The increase was non-significant in mice exposed only to 1% DSS (1.6-fold increase, P = 0.408) or 3% DSS (2.6-fold increase, P = 0.126), but reached statistical significance in mice receiving 3% DSS–1% DSS (3.6-fold increase, P = 0.015) (Table II). In mice exposed to DSS, repetitive WAS-induced corticosterone changes were attenuated and, although, in general, an increase in plasma corticosterone levels was observed associated to the stress, statistical significance was not reached for any of the experimental groups (Table II).

Table II.  Weight of thymus, spleen and adrenal glands and plasma corticosterone concentrations at necropsy (day 40).

Treatments		Thymus (mg)	Spleen (mg)	Adrenals (mg)	Corticosterone (ng/ml)
Water–water	Sham stress	58.2 ± 1.4	77.7 ± 4.3	6.9 ± 0.5	66.8 ± 8.7
	WAS	37.8 ± 3.4	55.2 ± 4.2	7.3 ± 0.6	225.9 ± 31.4^*
Water–1% DSS	Sham stress	51.7 ± 4.3	77.8 ± 7.5	4.6 ± 0.3	107.5 ± 25.7
	WAS	59.8 ± 8.1	60.1 ± 4.4	8.6 ± 0.6^*	193.7 ± 22.7
3% DSS–water	Sham stress	48.7 ± 2.3	78.4 ± 7.9	3.7 ± 0.1	176.6 ± 36.4
	WAS	42.1 ± 4.1	78.3 ± 7.2	5.9 ± 0.3^*	311.2 ± 36.6^†
3% DSS—1% DSS	Sham stress	49.1 ± 3.1	75.3 ± 7.6	4.9 ± 0.3	240.7 ± 39.7^‡
	WAS	56.5, 58.9^¶	89.4 ± 13.2	9.8 ± 0.9^*	141.78 ± 16.9

* P < 0.05 vs. respective sham stress group

^† P = 0.061 vs. respective sham stress group

^‡ P < 0.05 vs. water–sham stress–water. Data are group mean ± SEM (n = six mice per group)

^¶ Because of technical problems, the weight of the thymus was assessed only in two mice in this group (individual values are shown).

Basal levels of metabolic parameters (plasma glucose, triglycerides and cholesterol) in the water–sham stress–water group were: glucose, 12.52 ± 0.64 mM; triglycerides, 0.60 ± 0.04 mM; cholesterol, 1.95 ± 0.08 mM (mean ± SEM of six mice per group). Values for the same parameters were similar in magnitude across the different experimental groups, with no consistent treatment-related variation (data not shown).

Weight of body organs

At necropsy, weight of the adrenals was increased in all groups exposed to DSS (either 3 or 1%) plus WAS compared with their respective sham stress groups (Table II). However, no differences between the sham stress and WAS groups was observed in mice receiving normal water through the experiment. No consistent treatment-associated changes were observed for the weight of the spleen or the thymus (Table II).

Macroscopic assessment of inflammation

At day 40, the colonic inflammatory score was significantly higher in the experimental groups previously exposed to 3% DSS compared with those that received normal water during the same period (Figure 4B). In the 3% DSS-treated mice, the exposure to stress augmented the inflammatory score when compared to mice maintained in non-stressful conditions. This effect was significant in mice that, in addition to 3% DSS, were exposed to 1% DSS during the stress period (3% DSS–WAS–1% DSS: 3.2 ± 0.3; 3% DSS–sham stress–1% DSS: 2.0 ± 0.3; P < 0.05); while a clear tendency was observed in mice receiving normal water (3% DSS–WAS–water: 3.2 ± 0.6; 3% DSS–sham stress–water: 1.8 ± 0.8; P = 0.060) (Figure 4B). Macroscopic inflammatory scores changed in the same direction as the colon length and the relative colon weight (Figure 4C, D) (though differences in these parameters did not reach statistical significance). Exposure to 1% DSS and/or WAS in mice otherwise healthy, did not elicit any inflammatory response (Figure 4B–D).

Microscopic assessment of inflammation and colonic morphometry

Mice receiving water and subjected to either WAS or sham stress showed no signs of intestinal inflammation, with essentially normal histological features (Figure 5A, B). Similarly, mice receiving water and 1% DSS, with or without WAS, showed a normal colonic histology (data not shown). On the other hand, microscopic scores were significantly higher in those groups exposed to 3% DSS, irrespective of a further exposure to either water or 1% DSS, when compared with groups receiving water–water or water–1% DSS (Figure 6A). Histological evaluation of tissue sections from mice receiving 3% DSS–sham stress–water or 3% DSS–sham stress–1% DSS showed a low level of inflammation, reflected by mild infiltration of inflammatory cells in mucosa and submucosa, fully regenerated crypts and sparse focal ulcerations (Figure 5C, E). Addition of WAS to 3% DSS-exposed mice showed a clear tendency to worsen inflammation, although statistical significance was only reached between the 3% DSS–water groups but not between the 3% DSS–1% DSS groups (Figure 6A). In accordance, histological evaluation of tissue sections of mice exposed to 3% DSS–WAS, with or without 1% DSS, showed moderate infiltration of inflammatory cells within the mucosa and submucosa, irregular epithelial architecture due to epithelial regeneration, and in some cases extensive ulcerations (Figure 5D, F).

Figure 5 Representative haematoxylin and eosin-stained tissue sections from the different experimental groups. (A) Water–sham–water. (B) Water–WAS–water. (C) 3% DSS–sham–water. (D) 3% DSS–WAS–water. (E) 3% DSS–sham–1% DSS. (F) 3% DSS–WAS–1% DSS. The arrowheads in D and F denote areas of epithelial damage. Scale bar: 50 μm.

Figure 6 Colonic microscopic inflammatory score and morphometric analysis of the colon at the time of necropsy. (A) Microscopic scores, on a scale of 0–5. *P < 0.05 vs. water + water or water+1% DSS groups (B) Morphometic analysis of the total colonic wall showing the thickness of the mucosa-submucosa (hatched bars) and the muscle layers (empty bar). *P < 0.05 vs. respective sham stress group (ANOVA). Data in mean ± SEM values for six mice per group.

The morphometric analysis of the colon showed changes in the same direction as the microscopic inflammatory score. In mice exposed to 3% DSS, irrespective of a further exposure to either water or 1% DSS, the total thickness of the colonic wall was increased when compared with mice receiving normal water, although statistical significance was not reached. However, the addition of stress (with or without exposure to 1% DSS) resulted in a significant increase in thickness (Figure 6B). Changes in the total wall thickness corresponded to increases in the thickness of both the muscle layers and the mucosa–submucosa compartment.

Local markers of inflammation

In general, colonic tissue concentrations of inflammatory cyto- and chemokines motifs followed the pattern described for the macro- and microscopic assessment of inflammation. In all experimental groups exposed to normal water from day 0 to day 5, independently of the treatments received, a posteriori (WAS or sham stress and water or 1% DSS), the content of cytokines and C-X-C motifs was low, in some cases at the limit of the detection level (Figure 7). Similarly, mice exposed to 3% DSS followed by normal water had only moderately elevated levels of cyto- and chemokines motifs. Exposure to stress increased the levels of IL-12p40 and CCL5, with little or no effect on other markers (Figure 7). Interestingly, exposure to 1% DSS, with or without stress, tended to accentuate the changes in some of the markers evaluated: particularly IL-1β, IL-6, CXCL1 and CCL2 (Figure 7).

Figure 7 Colonic concentrations of inflammatory markers in the different experimental groups. The open bars represent the sham stress control groups and the black bars the water avoidance stress groups. IL-1β, interleukin-1β; IL12p40, interleukin 12p40; CCL5, regulated upon activation, normal T-cell expressed, and presumably secreted-RANTES/CCL5 (CCL5); CXCL1, chemokine (C-X-C motif) ligand 1. Data are mean ± SEM values for six mice per group. *P < 0.05 vs. respective sham stress group (ANOVA).

At the gene transcription level, the expression of IL-6 and COX-2 mRNA was significantly elevated in the 3% DSS–WAS–1% DSS-treated mice compared to their sham stress counterparts (Figure 8). No significant differences were found in iNOS, TNF-α or IFN-γ mRNA expression in the 3% DSS-treated animals with or without stress (Figure 8 and data not shown). Healthy mice exposed to the combination of WAS and 1% DSS showed a mild elevation in IL-6, TNF-α and iNOS mRNA expression, whereas healthy mice exposed to either WAS or 1% DSS alone showed no changes (data not shown).

Figure 8 Colonic expression of inflammatory markers: IL-6, TNF, iNOS and COX-2 in the different experimental groups. The open bars represent the sham stress control groups and the black bars the water avoidance stress groups. IL-6, Interleukin 6; TNF, tumour necrosis factor α; iNOS, inducible form of the nitric oxide synthase; COX-2, cyclooxygense 2. Data are mean ± SEM values for six mice per group. *P < 0.05 vs. respective sham stress group (ANOVA).

Systemic markers of inflammation

Plasma concentrations of the acute-phase protein haptoglobin were, in general, low in all the experimental groups, with no differences among them (Figure 9). Nevertheless, haptoglobin levels tended to be increased in the 3% DSS–WAS–1% DSS and 3% DSS–WAS–water groups (levels varying between 1.65 and 0.35 g/l), while in the rest of the experimental groups the levels varied between 0.05 and 0.24 g/l. Similarly, plasma concentrations of IL-6 and CXCL1 were in general low (for IL-6, under the detection level in most cases) with significant changes only in the 3% DSS–WAS–1% DSS group (Figure 9). No consistent treatment-associated changes were observed for the plasma levels of IL-12p40 (data not shown).

Figure 9 Plasma concentrations of inflammatory markers (haptoglobin, IL-6 and CXCL1) in the different experimental groups. The open bars represent the sham stress control groups and the black bars the water avoidance stress groups. IL-6, interleukin 6; CXCL1, chemokine (C-X-C motif) ligand 1; DL, at the limit of detection. Data are mean ± SEM values for six mice per group. *P < 0.05 vs. respective sham stress group (ANOVA).

Multivariate analysis

Datasets corresponding to 31 variables (loadings) from 48 mice were analysed by PCA. The resulting model included three principal components (PC), with PC1 displaying the degree of inflammation, PC2 the response to stress and PC3 a combination of both. Mice exposed to WAS showed movement in PC2 in an inverse direction compared to the disease direction (PC1). When the PC loadings were inspected, markers characterizing the stress response (mainly acute body weight loss during WAS and mean and cumulative pellet output) correlated well for healthy mice exposed to WAS. Similarly, markers characterizing the inflammatory response (local inflammatory markers, inflammatory score and histology) correlated for mice exposed to 3% DSS–WAS–1% DSS. Mice exposed to 3% DSS–WAS–water or 3% DSS–sham–1% DSS moved towards the third component. Interestingly, when we compared healthy mice exposed to WAS–1% DSS (water–WAS–1% DSS group) to healthy mice exposed to sham stress–water (water–sham stress–water group), the main parameters defining this experimental group were local inflammatory markers, mean and cumulative pellet output and plasma corticosterone levels. However, a similar comparison between the water–sham–1% DSS and the water–sham stress–water groups showed no association to either of these markers.

Discussion

The present study shows that, in mice with experimentally induced chronic colitis, repetitive psychological stress reactivated the inflammatory state, an effect that was enhanced by the addition of an innocuous concentration of DSS (1%). The results obtained suggest that the stress, per se, enhances macroscopic and microscopic indices of colonic inflammation, with little effects on biochemical markers of inflammation. Interestingly, the concomitant presence of a sub-threshold inflammatory stimulus, that per se did not induce inflammation, resulted in a more consistent induction of inflammatory markers.

Several environmental factors, such as smoking or psychological stress, have been suggested as triggers of IBD in humans (Mawdsley and Rampton 2005; Mawdsley et al. 2006). Clinical observations provided anecdotal evidence linking the relapse of the disease with the occurrence of stressful life events. However, the role of stress in the pathophysiology of IBD remains controversial. Recent reports suggested that stressful life events or depression are risk factors for the onset or relapse in CD patients (Mardini et al. 2004; Lerebours et al. 2007). Similarly, IBD patients in remission exposed to 5 days repeated stress (cold pressor test) showed changes in stress hormones, activation of mast cells and epithelial cell damage when compared with healthy controls (Farhadi et al. 2005). All together, these observations further support a significant role for stress in the recurrence of inflammation in IBD.

DSS-induced colitis in mice is a valid experimental model to study the pathogenic mechanisms underlying IBD (Okayasu et al. 1990; Melgar et al. 2005, 2006). Besides the characteristic generation of acute inflammatory responses with translational value to the active phase of inflammation present in IBD, this acute response evolves to a mild chronic inflammatory condition. This chronic inflammation is characterized by infiltration of inflammatory cells, production of inflammatory mediators and irregular epithelial structure in the colon, with few clinical or systemic signs of inflammation, similar to the quiescent phase of the disease in humans (Melgar et al. 2005, 2006). Moreover, we show here that this chronic colonic inflammation can be reactivated by stress, reminiscent of the relapse of the disease observed in humans. In particular, under the present experimental conditions, stress increased the macroscopic and microscopic indices of inflammation and increased, or showed a clear trend towards increased, several markers associated to gut inflammatory conditions. The changes in IL-1, IL-6 and CXCL1 (murine IL-8 homolog, with the main function to attract neutrophils) are in line with the cytokine response found in acute inflammatory responses in UC patients with active disease and in mice with DSS-induced acute colitis (Hirano 1996; Guimbaud et al. 1998; Okayasu et al. 1990; Melgar et al. 2005). IL-1 and IL-6 belong to the acute-phase proteins group, which are highly activated upon different types of trauma (e.g. stress or sepsis) and affect several immunomodulatory functions, such as the activation of the HPA axis (Dunn 2000). In addition, these cytokines have been suggested to play an important role in the initiation and perpetuation of colitis (Guimbaud et al. 1998; Podolsky 2002). Thus, elevation of levels of these cytokines in mice exposed to 3% DSS–WAS–1% DSS further supports a reactivating role of stress on chronic colitis.

Whether or not other factors likely to influence the course of IBD are also able to reactivate DSS-induced chronic inflammation deserves further study. Therefore, DSS-induced intestinal inflammation might be a valid animal model mimicking all the aspects of IBD, from the active phase of inflammation to the reactivation of the disease by specific triggering factors, such as stress.

Acute and chronic WAS has been extensively used in rats to asses the effects of stress on gastrointestinal functions (Martinez et al. 1997; Yang et al. 2006). However, the model has not been used in a systematic manner in mice (Cameron and Perdue 2005). Results obtained show that exposure to WAS stimulates colonic propulsive motor activity, leading to increased defecation during the time of stress, as observed in other models of stress in mice (Martinez et al. 2004), and induces a transient body weight lost (by 3% of the pre-stress weight), while mice maintained in non-stressful conditions showed no major changes in these parameters. The increased defecation rate, representing only about 50% of the body weight difference pre- and post-stress, cannot solely explain the body weight loss. Therefore, acute metabolic adjustments resulting in a transient loss of body weight may also operate during the stress. However, no variations in the plasma levels of metabolic markers were observed immediately after the last stress session. In addition, responses to stress were stable across the 7-day period of repetitive WAS, indicating that the mice do not habituate to the stress protocol. Furthermore, in the chronically stressed mice, the adrenal glands were enlarged, the plasma levels of corticosterone were increased and colonic faecal content was softer when compared to non-stressed mice. The latter might be associated with stress-induced colonic secretory responses, as described previously (Saunders et al. 1994; Santos et al. 1998). All together, these observations show that repetitive WAS is a valid chronic stressor in mice, eliciting consistent and reproducible functional changes, and can be used as a model of repetitive psychological stress in this species.

Interestingly, stress-induced faecal pellet output, although still significantly increased, was lower in mice receiving DSS, regardless of the concentration used, when compared with the responses observed in healthy stressed mice. This might suggest that the presence of a chronic inflammation, or even the early phase of a novel inflammatory response (likely associated to the exposure to 1% DSS in otherwise healthy mice), might affect colonic motility resulting in an impairment of colonic propulsive activity. It is generally recognized that abnormal motility is one of the consequences of intestinal inflammation, and that this abnormality can lead to constipation in IBD (Ozaki et al. 2005). Similarly, reduced colonic smooth muscle contractility has been demonstrated in rats with trinitrobenzene sulfonic acid (TNBS)- or DSS-induced colitis (Kinoshita et al. 2003; Kiyosue et al. 2006; Sato et al. 2006), an effect likely associated with the presence of pro-inflammatory cytokines (Ozaki et al. 2005). In line with this hypothesis, the degree of impairment of the colonic motor responses to stress was graded and correlated well with the extent of the inflammatory responses observed and the local levels of inflammatory markers. Moreover, similar to the observations here during chronic inflammation, reduced defecatory responses during acute stress were also observed in rats with TNBS-induced acute colitis (Kresse et al. 2001), further supporting an interaction between colonic inflammation and the motor response to stress.

In mice exposed to water–water or water–1% DSS, irrespective of the stress treatment, the clinical parameters and the local and systemic levels of inflammatory markers were low and basically within the margins characteristics of normal, healthy mice. This correlated well with the absence of histological alterations and shows that, in the present experimental conditions, exposure to 1% DSS or stress, alone or in combination, in otherwise healthy mice, was unable to elicit an inflammatory response.

In non-stressed, 3% DSS-treated mice, histological evaluation of colonic tissues showed mild histopathological changes and overall low values for local inflammatory markers, suggesting the presence of a mild chronic inflammatory condition, similar to that observed in humans during the latent phases of IBD (Mawdsley et al. 2006; Ishiguro 1999) and to our observations in this model (unpublished observations). Mice with chronic inflammation subjected to repetitive WAS showed macroscopic signs of inflammatory reactivation, which was corroborated by an increased histological microscopic score and increased levels of local inflammatory markers, when compared to their respective sham stress group. This suggests that stress reactivates a latent inflammatory condition but, per se, does not induce inflammation, as shown by the lack of inflammation in mice exposed solely to WAS. These observations contrast with previous data in mice showing that repeated psychological stress, without any other facilitating factor, might lead to the development of intestinal inflammation (Reber et al. 2007). However, significant differences between this report and our study should be mentioned, including the duration (19 vs. 7 days in the present study) and the intensity of the stressor (continuous vs. intermittent stress, 1 h/day, in the present study). Our observations are, however, in agreement with recent reports showing that stress, together with a low concentration of an inflammation-inducing agent, aggravated acute inflammatory responses (as determined by myeloperoxidase content and histology) in rats and mice (Qiu et al. 1999; Saunders et al. 2006), and suggest that similar effects can be seen during the chronic phase of inflammation. Similarly, a recent study in patients with inactive UC shows that acute psychological stress (a modified dichotomous listening test) induced systemic and mucosal pro-inflammatory responses (as determined by changes in the levels of TNF-α, IL-6 and reactive oxygen metabolites in colonic mucosa), which could contribute to the exacerbation of UC (Mawdsley et al. 2006). Interestingly, UC patients with histological signs of inflammation before the stress showed an enhanced stress response (Mawdsley et al. 2006). Together with our observations in mice, these data in humans strongly suggest that the stressful events may act, per se, as pro-inflammatory factors.

The reactivating effects of stress may be associated with the activation and/or the infiltration of inflammatory cells, as suggested by the elevation in the colonic content of chemo- and cytokines involved either in directing T cells (CCL5) and neutrophils (CXCL1) to the tissue or in activating resident immune cells (IL-1β, IL-6, IL-12p40). Interestingly, CD4⁺T cells are a prerequisite for the reactivation of dinitrobenzene sulfonic acid-induced colitis in mice (Qiu et al. 1999). Stress may also influence the intestinal barrier function, for example increasing mucosal permeability, thus affecting bacterial–host interactions and the transport of luminal macromolecules (Cameron and Perdue 2005; Keita et al. 2006; Cario et al. 2007), an effect dependent on cholinergic mechanisms (Saunders et al. 1997; Gareau et al. 2007) and activation of mast cells (Yu and Perdue 2001; Soderholm et al. 2002; Jacob et al. 2005; Demaude et al. 2006). The exact mechanisms mediating chronic WAS-induced reactivation of DSS-induced chronic inflammation in the present experimental conditions need to be further characterized by taking into consideration these potential pathways.

An apparent mismatch between the weight of the adrenal glands and the plasma levels of corticosterone was observed in mice exposed to DSS (either 3 or 1%). Corticosterone levels in non-stressed DSS-exposed mice were moderately increased when compared with healthy mice in non-stressful conditions, but blunted in response to the repetitive stress, as described previously (Million et al. 1999; Kresse et al. 2001; Edgar et al. 2003; Silberman et al. 2003; Reber et al. 2006). Despite these changes, the adrenal weight was only increased in mice exposed to DSS and subjected to stress. These observations confirm a significant interaction between the neuroendocrine response to stress and inflammation, as previously described in other animal models (Kresse et al. 2001; Reber et al. 2006, 2007) and might be associated with a direct effect of pro-inflammatory cytokines on the adrenal gland (Bornstein et al. 2004), resulting in an increase in glandular weight and in a loss of functional responsiveness to the activation of the hypothalamic-pituitary-adrenal, in turn leading to blunted corticosterone responses during stress (Edgar et al. 2003; Silberman et al. 2003; Reber et al. 2006). Moreover, TNB-induced acute colitis in rats has been shown to reduce CRF gene expression in the hypothalamus, leading to dampened corticosterone responses to environmental acute stressors (Kresse et al. 2001). A similar mechanism might also be operating during chronic inflammation, contributing to the reduced endocrine response to repetitive stress.

In summary, the current studies show, for the first time, that repetitive psychological stress is able to reactivate a quiescent chronic colonic inflammation. Moreover, stress-induced inflammatory responses were enhanced by the presence of a sub-threshold inflammatory stimulus (1% DSS), while the same stimulus was ineffective per se. Similarly, the results obtained question the ability of stress to induce, per se, inflammatory responses within the gut in the absence of a latent predisposing inflammatory condition. Nevertheless, further studies should address potential effects associated with the intensity and the duration of the stressor used. Furthermore, the current protocol showed that repetitive WAS is a valid mild stressor in mice and can be used to characterize the functional effects of stress in this species. Thus, the data presented support the view that stress should be regarded as an environmental factor with the capability to trigger a relapse of IBD in humans.

Acknowledgements

We gratefully acknowledge Laboratory Animal Resources (AstraZeneca R&D Mölndal) for their support with animal care and the Analytical Biochemistry group (AstraZeneca R&D Mölndal) for bioanalytical analysis.