Genes and (2000) 1, 213–218  2000 Macmillan Publishers Ltd All rights reserved 1466-4879/00 $15.00 www.nature.com/gene Eosinophilia is attenuated in experimental colitis induced in IL-5 deficient mice

L Stevceva1, P Pavli1, A Husband2, KI Matthaei3, IG Young3 and WF Doe1 1Division of Molecular Medicine, and 3Division of Biochemistry and Molecular Biology, John Curtin School of Medical Research, ANU, Canberra, ACT, 2601; 2Department of Veterinary Anatomy and Pathology, University of Sydney, NSW, 2006, Australia

Tissue eosinophilia is a feature of idiopathic inflammatory bowel disease and other forms of colonic inflammation but it is not clear whether the role of in the disease process is to contribute to tissue damage. 5 (IL-5) stimulates production and activation of eosinophils in vitro and enhances immunoglobulin A (IgA) production. As very little is known about the function of IL-5 in the colon, the aim of this study was to assess its role in colonic inflammation. IL-5 deficient mice were studied using the dextran sulphate sodium (DSS)-induced colitis model and the results compared to a congenic IL-5+/+ strain. The absence of IL-5 resulted in reduction of tissue eosinophilia (P Ͻ 0.0001) but was not reflected in differences in the severity of the disease (P Ͼ 0.5) or in the extent of tissue damage in this model of colitis. Numbers of immunoglobulin-containing cells in IL-5 deficient mice were similar to those in the IL-5+ mice. We conclude that the main role of IL-5 in DSS-induced colonic inflammation is to attract a population of eosinophils which do not appear to contribute significantly to the initiation or development of tissue damage in this model of colitis. Genes and Immunity (2000) 1, 213– 218.

Keywords: experimental colitis; dextran sulphate sodium; interleukin 5

Introduction known, in this paper we explore the function of IL-5 in a mouse model of colitis using IL-5 deficient mice. Our Ulcerative colitis is a disease of unknown aetiology findings suggest a role for IL-5 in recruitment characterised by a chronic relapsing course. The disease in acute colonic inflammation but indicate that IL-5 can be induced experimentally in mice by dextran sul- deficiency does not affect the course of the disease or the 1 phate sodium (DSS) given in the drinking water. Our number of immunoglobulin-producing cells in the studies using this model have confirmed that the histo- mucosa. pathological changes that are induced by this method resemble ulcerative colitis. Infiltration by eosinophils into the inflamed areas of Results the colon is a feature of both the clinical disease and the The results described below comprise a summary of a model of DSS-induced colitis (see below). The eosinophil total of five experiments. lineage appears to be specifically regulated by interleukin 5 (IL-5), a produced by the T helper 2 subset IL-5 influences eosinophil infiltration 2 (Th2) as well as by eosinophils and mast cells. Although Baseline eosinophil numbers in the colon of healthy IL-5 earlier studies suggested additional activities of IL-5 in deficient mice varied from 0 to an average of 0.25 cells 3 regulating mouse B cells and production recent per transverse section of the colon and were significantly studies with an IL-5 deficient mouse strain indicate that lower (P = 0.0001) than in the congenic strain (2.08 to 3.6). the regulation of eosinophilia may be the only obligatory Eosinophil numbers increased in both strains during the 4 role for IL-5 in the adult mouse. IL-5 deficient mice, pro- course of the DSS-induced colitis. Administration of 5% 4 duced by gene targeting of embryonal stem cells, appear DSS was associated with lower eosinophil counts in IL-5 to be normal in appearance and do not develop any obvi- deficient mice compared to the congenic strain, a differ- ous disease. Tissue eosinophilia induced by parasitic ence which was not statistically significant (P = 0.079). infections is abolished in these mice although baseline Eosinophils were very low or absent in IL-5 deficient numbers of eosinophils are close to normal. Since the role mice exposed to 2.5% DSS (Figure 1). The 2.5% DSS of IL-5 in the mucosal lesion of ulcerative colitis is not induced a significantly higher eosinophil count at day 9 in the IL-5+/+ mice (P = 0.0001) compared to the congenic IL-5 deficient strain (Figure 1). Representative data in Correspondence: Dr Paul Pavli, Gastroenterology Unit, The Canberra Figure 1 show the number of eosinophils at different time Hospital, P.O. Box 11, Woden ACT 2606, Australia points during the course of the disease. The absence of Parts of this work have been presented at the 95th Annual Meeting IL-5 did not completely suppress the eosinophil infil- of the American Gastroenterological Association and published in abstract form in Gastroenterology 1995; 108: A922 tration into the colon suggesting that a minor proportion Received 25 June 1999; revised 9 August 1999; accepted 3 Sep- of the infiltrating eosinophil population was not IL-5 tember 1999 dependent. Eosinophilic infiltration in the colon was not Eosinophilia and IL-5 deficient mice L Stevceva et al 214

Figure 1 Eosinophil counts in the colon. Eosinophil counts were performed in IL-5 deficient mice and the congenic IL-5+/+ strain before and on 4, 7 and 9 days after exposure to DSS. Similar results were obtained in five consecutive experiments. (a) Eosinophil counts in mice exposed to 2.5% DSS; (b) eosinophil counts in mice exposed to 5% DSS.

provoked by the high numbers of and their and rectum. The first signs of inflammation are apparent products. numbers ranged from 2.60 ± 0.77 after 3 days exposure to DSS and present clinically as (mean ± standard deviation) in the 2.5% DSS-induced progressive weight loss. Subsequently, mice develop dia- colitis and 2.75 ± 1.23 in the 5% DSS-induced colitis in rrhoea and rectal bleeding. After ingestion of 2.5% or 5% the IL-5+/+ strain, to 3.62 ± 1.42 in the 2.5% DSS-induced DSS the course of the disease in IL-5 deficient mice was colitis and 2.63 ± 1.13 in the 5% DSS-induced colitis in no different to that seen in the congenic IL-5+/+ strain the IL-5 deficient mice indicating that there was no (Figure 2). significant difference between IL-5 deficient mice and the Histological examination did not show any significant congenic strain. differences in the distribution of the colitis or in the severity of tissue damage. Ulceration developed in both IL-5 deficiency does not affect the severity of the mouse strains to a similar extent. The histological score, colitis a quantitative measure of the severity of tissue damage Dextran sulphate sodium-induced colitis develops as an was no different in IL-5 deficient mice (11.25 ± 0.96) com- acute inflammation affecting predominantly the left colon pared to IL-5+/+ control mice (10.60 ± 1.67, P = 0.4).

Genes and Immunity Eosinophilia and IL-5 deficient mice L Stevceva et al 215 ulin A (IgA)+, (IgM)+ and immuno- globulin G (IgG)+ cells were similar in healthy IL-5 deficient mice and their congenic controls. Immunoglob- ulin E (IgE)-containing cells were undetectable in the colon in both mouse strains. Colitis induced by 2.5% DSS was characterised by an increase in total immunoglob- ulin-containing cell numbers (Table 1) most of which could be accounted for by an increase in IgA+ cells. There was no significant difference between the numbers of IgA+, IgM+ or IgG+-containing cells in the inflamed colon of IL-5 deficient mice and the congenic IL-5+/+ strain (Table 1). By contrast, colitis induced by 5% DSS resulted in a significant reduction (P = 0.04) of the numbers of total immunoglobulin-containing cells as well as in reduced expression of IgA and IgG-expressing cells. The table shows the decrease in the number of immunoglob- ulin-containing cells at day 8 of the disease compared to the healthy colon before the administration of dextran. The reason for the difference in Ig-containing cell num- bers when two concentrations of DSS are used is unclear.

Discussion The results described in this study suggest that the main role of IL-5 in colonic inflammation is to promote the infiltration of eosinophils into the colon. The absence of changes in the population of immunoglobulin-containing cells indicates that IL-5 may not have an obligatory role in mucosal IgA and IgG production in vivo. This contrasts with the effects of IL-5 previously observed ‘in vitro’.5–7 In mice, the main activities of IL-5 involve eosinophils and B cells whereas in humans, IL-5 principally affects the eosinophil lineage but its effects on B cells remain controversial. IL-5 does not appear to influence small resting cells but stimulates large pre-activated B cells, inducing proliferation and secretion of immunoglobulins in vitro.5 IL-5 also enhances secretion of IgA by precom- mitted B cells6 and affects IgG production by - primed B cells.7 In addition, IL-5 stimulates the pro- duction of eosinophils in cultures of human bone mar- row,8 acts as a chemo-attractant for mature murine eosinophils and induces superoxide anion production.9 Much less however is known about the activity of IL-5 in vivo. This paper explores the role of this cytokine in the DSS-induced mouse model of acute ulcerative colitis. IL-5 did not influence immunoglobulin-containing cell numbers in the intestine during the course of DSS- induced colitis nor does it appear to be necessary for anti- Figure 2 Activity of the disease during the course of DSS-induced body production. These findings confirm findings by colitis. Activity of the disease was measured by the disease activity 4 index at day 4, 7 and 9 of the disease. The severity of the colitis Kopf et al that IL-5 does not have an obligatory role in was not significantly different in the IL-5 deficient mice compared either growth or antibody production in vivo. The to the IL-5+ congenic strain regardless of the concentration of DSS levels of colonic eosinophils in healthy IL-5 deficient mice used to induce colitis (P = 0.6 and P = 0.62). (a) Disease activity were lower than in controls and despite an increase dur- index in mice exposed to 2.5% DSS (n = 25); (b) disease activity = ing the course of the disease there was a significant index in mice exposed to 5% DSS (n 24). reduction in eosinophil numbers infiltrating the colon at day 9 in animals exposed to 2.5% DSS in the absence of These findings indicate that IL-5 does not influence IL-5. These findings indicate that IL-5 contributes to the significantly the severity or the outcome of colonic colonic inflammatory lesion mainly via its action on inflammation in this model. eosinophils. However, the attenuation of eosinophil infil- tration resulting from the absence of IL-5 does not result IL-5 does not influence immunoglobulin-containing in a corresponding change in the extent or severity of cell numbers in the mucosa tissue damage, raising questions about the effects of the Immunofluorescent staining for immunoglobulin- eosinophil in inflammation. Similarly, the absence of IL-5 containing cells in the colonic mucosa was performed in appeared to have no effect on the numbers of immuno- both mouse strains. Baseline numbers of immunoglob- globulin-containing cells indicating that IL-5 is not essen-

Genes and Immunity Eosinophilia and IL-5 deficient mice L Stevceva et al 216 Table 1 Number of immunoglobulin-producing cells/cm of the colon. The values shown are mean values ± standard deviation

Positive Untreated mice P value Day 8, 2.5% DSS P value Day 8, 5% DSS P value cells/cm IL-5+/+ IL-5−/− IL-5+/+ IL-5−/− IL-5+/+ IL-5−/−

IgA 517.9 ± 145.6 694.4 ± 267.0 0.25 1102.0 ± 593.0 877.5 ± 782.3 0.58 276.2 ± 116.6 317.0 ± 173.4 0.47 IgG 47.0 ± 62.0 45.5 ± 40.2 0.46 35.3 ± 18.8 43.5 ± 21.7 0.72 14.3 ± 5.3 67.7 ± 104.4 0.65 IgM 48.3 ± 36.7 42.8 ± 22.5 0.75 59.5 ± 51.7 18.4 ± 14.7 0.03 24.0 ± 18.2 39.9 ± 36.4 0.72 IgE nil nil NA nil nil NA nil nil NA

tial to the regulation of IgA and IgG production in vivo Dextran sulphate sodium in the inflamed intestinal mucosa. The dextran sulphate sodium used was a sodium salt of Increased numbers of mucosal eosinophils in inflam- molecular weight of 41 K and with sulphur substitution matory bowel disease (IBD) have been reported in several of 16.3 (TdB Consultancy, Uppsala, Sweden). DSS was studies. While the mucosal eosinophil count is higher given as 2.5% and 5% solutions in distilled water for 8 than in normal control subjects in cases of long standing days. ulcerative colitis, in a severe first attack of ulcerative col- itis, the number of eosinophils is not significantly higher Disease activity index from control levels.10 It has been suggested that the role Disease activity index (DAI) was derived from the three of eosinophils in ulcerative colitis may be inhibitory or major clinical signs (weight loss, diarrhoea and rectal regulatory.11 Heatley and James12 have shown that a high bleeding) as previously described.16 The clinical signs eosinophil count in the rectal biopsy indicates a relatively were analysed separately and correlated with the histo- good prognosis, the colitis being self-limiting or easily logical score. The final formula for the DAI was evaluated controlled with medical therapy. by correlation with the histological score. The DAI was With regard to animal models of ulcerative colitis, high defined as: eosinophil counts are present in carageenan-induced col- DAI = (weight loss) + (diarrhoea score) + (rectal bleed- 13 itis and in the present studies we demonstrated ing score). increased numbers of eosinophils in the inflammatory infiltrate in DSS-induced colitis. The numbers of eosino- Body weight loss: Body weight loss was calculated as the phils are much higher if the colitis is induced by a lower difference between the predicted body weight and the concentration of DSS although the significance of this actual body weight on a particular day. The formula for finding is unclear. Macrophages and their products are predicted body weight was derived by simple regression important factors in the recruitment of eosinophils14,15 using the body weight data for the control group. The but we did not demonstrate any difference in numbers of following formula was derived: Y = a + kx, where macrophages in the inflammatory infiltrate between IL-5 Y = body weight change (loss or gain), k = daily increase deficient mice and controls. Our results, however indicate in body weight, x = day, a = starting body weight. The that the recruitment of eosinophils to the inflamed region daily increase in body weight (k) was 0.251 for the IL- is influenced by the activity of IL-5 but that the reduced 5+/+ strain and 0.242 for the IL-5 deficient mouse strain. eosinophil responses in IL-5 deficient mice did not alter the severity of the disease or the extent of tissue damage. Diarrhoea: The appearance of diarrhoea was defined as These findings suggest that neither the eosinophil nor IL-5 /faecal material adherent to anal fur. The presence contributes significantly to tissue damage nor does it or absence of diarrhoea was scored as either 1 or 0 induce a protective immunological response in the respectively. The cumulative score for diarrhoea was cal- inflamed colon. While their exact roles is yet to be culated by adding the score for each day and dividing determined this study deals with the acute phase of DSS- this number by the number of days of exposure. induced colitis some of which may not be mediated immunologically. Further studies are needed to deter- Rectal bleeding: The appearance of rectal bleeding was mine the impact of eosinophils and IL-5 in more chronic defined as diarrhoea containing blood or mucus or frank phases of inflammation. rectal bleeding. The final score for rectal bleeding was calculated as described for diarrhoea.

Materials and methods Histopathological examination and scoring Tissue specimens were fixed in 10% neutral formalin and Mice embedded in paraffin. Colons were divided into right, IL-5 deficient mice were generated in C57Bl/6 stem cells transverse and left and specimens stained with haema- by gene targeting and were regularly checked for IL-5 toxylin and eosin. Histological examination was perfor- deficiency.4 Homozygous animals were obtained by med on serial sections. The most affected part of the breeding of heterozygous parents and were identified by colon was scored blindly using a standard 12-point histo- polymerase chain reaction analysis. Mice were fed sterile logical scoring system.17 food and water ad libitum and were regularly checked for viruses and parasites. The studies were approved by The Eosinophil counts Animal Experimentation Ethics Committee of The Aus- Eosinophil counts were performed on serial, Giemsa- tralian National University. stained, 4 ␮ thick, transverse sections of the left colon and

Genes and Immunity Eosinophilia and IL-5 deficient mice L Stevceva et al 217 were expressed as the average eosinophil count per full antigen. Di-aminobenzidine (DAB, 1:5 in PBS combined ␮ transverse section of the colon. The results were with 1 ml imidazole and 15 lH2O2) was used as a sub- confirmed by additional counts per high power field strate and was applied for 10 min following the 30 min (×250 magnification) and by counting the numbers of incubation in ABC and 2 × 5 min. washing in PBS. Tissue eosinophils per cm2 of using video sections were washed in water for 3 min and coun- image analysis. terstained using haematoxylin.

Immunofluorescent staining Counting of positive cells: F 4/80 positive cells were coun- The specimens were fixed using cold ethanol,18 embed- ted using the NIH 1.55 video image analysis program ded in paraffin and 5 ␮ sections cut on to gelatine-coated and were expressed as the number of positive cells per slides. Immunoglobulin-containing cells were detected mm2 lamina propria. Only the lamina propria was by immunofluorescence as previously described.19 marked and its surface area measured. Crypts and epi- Briefly, tissues were dewaxed, rehydrated and then incu- thelial cells were not included in this calculation. The bated with goat anti-mouse IgA (␣ chain specific) results of the area measurement were corrected for the (SeraLab, Crawley Down, Sussex, UK), followed by tetra- magnification of the microscope and the screen. methylrhodamine-isothiocyanate (TRITC)-conjugated rabbit anti-goat immunoglobulin (Atlantic , References Scarborough, ME, USA) for detection of IgA-containing cells; fluorescein isothiocyanate (FITC) conjugated goat 1 Okayasu I, Hatakeyama S, Yamada M, Ohikusa T, Ingaki Y, anti-mouse IgM (␮ chain specific) (Cappel, Durham, NC, Nakaya RA. A novel method in the induction of reliable experi- USA) for detection of IgM-containing cells; FITC-conju- mental acute and chronic ulcerative colitis in mice. Gastroenterol- gated mouse against mouse IgG (␥- ogy 1990; 98: 694–702. 2 Dubucquoi S, Desreumaux P, Janin J et al. Interleukin 5 synthesis chain specific) (SeraLab); or FITC-conjugated mouse ε by eosinophils: association with granules and immunoglobulin- monoclonal antibody against mouse IgE ( -chain specific) dependent secretion. J Exp Med 1994; 179: 703–708. (SeraLab). Following each 20-min incubation, excess 3 Koike M, Takatsu K. IL-5 and its receptor: which role do they reagent was removed and the slides rinsed in phosphate play in the immune response? Int Arch Immunol 1994; buffered saline (PBS). Finally slides were rinsed for 1 h 104: 1–9. in PBS and mounted using 90% glycerol in PBS. The 4 Kopf K, Brombacher F, Hodgkin PD et al. IL-5 deficient mice number of Ig-containing cells in the intestine were coun- have a developmental defect in CD5+ B-1 cells and lack eosino- ted using a 40× objective and 10× eyepieces using a Zeiss philia but have normal antibody and cytotoxic responses. Axioskop 20 microscope with incident light illumination. Immunity 1996; 4: 15–24. 5 Harriman GR, Strober W. The immunology of interleukin 5. Year Each section of the intestine was scanned from the tips Immunol 1989; 5: 160–177. of the villi to the base of the mucosa, and cells were coun- 6 Ramsay AJ, Kohonen-Corish M. Interleukin-5 expressed by a ted in a total of 30 such scans. The average number of recombinant virus vector enhances specific mucosal IgA cells per scan was expressed as cells per linear cm of responses in vivo. Eur J Immunol 1993; 23: 3141–3145. intestine by multiplying by 30.3, a correction factor 7 Takatsu K, Tominaga A, Harada N et al. T cell replacing factor determined from the field width of 330 ␮m, as pre- (TRF)/interleukin 5 (IL-5): molecular and functional properties. viously described.20 Immunol Rev 1988; 102: 107–135. 8 Campbell HD, Tucker WJQ, Hort Y et al. Molecular cloning and Counting of macrophages expression of the gene encoding human eosinophil differen- tiation factor (IL-5). Proc Nat Acad Sci USA 1987; 84: 6629–6633. 9 Yamaguchi Y, Suda T, Suda J et al. Purified interleukin 5 sup- Immunocytochemistry: Periodate-lysine-paraformaldehyde ports the terminal differentiation and proliferation of murine (PLP)-fixed tissues were embedded in OCT compound eosinophilic precursors. J Exp Med 1988; 167: 43–56. (Miles Inc, Diagnostic Division, Pymble, NSW, Australia). 10 Willoughby CP, Piris J, Truelove SC. Tissue eosinophils in ulcer- Transverse, serial, 6 ␮ thick sections of the colon were ative colitis. Scand J Gastroent 1979; 14: 395–399. washed in PBS 2 × 5 min to remove OCT. Non-specific Fc 11 Beeson PB, Bass DA. Major Problems in Internal Medicine. No. binding was blocked by incubating with goat anti-mouse 14:1-269. The eosinophil. WB Saunders & Co: Philadelphia, Lon- IgG (H+L) (Caltag Immunodiagnostics, St Peters, NSW, don & Toronto, 1977. Australia) 100 ␮l/10 ml PBS and 2% sheep serum for 12 James PD, Heatley RV. Eosinophils in rectal mucosa. A simple 30 min. The macrophage-specific monoclonal antibody method of predicting the outcome of ulcerative proctocolitis? Gut 1979; 20: 787–791. F4/80 (1:3 in PBS)21 was used as a primary antibody and 13 Al-Suhail AA, Reid PE, Culling CF, Dunn WL, Clay MG. Studies applied after the blocking step. No primary antibody and of the degraded carrageenan-induced colitis of rabbits. II. control IgG2b (Pharmingen, Lane Cove, NSW, Changes in the epithelial glycoprotein O-acylated sialic acids Australia) were used as negative controls. Following 60 associated with the induction and healing phases. Histochem J min incubation at room temperature, slides were rinsed 1984; 16: 555–564. in PBS 2 × 5 min. Endogenous peroxidase activity was 14 Oliveira SH, Faccioli LH, Cunha FQ, Ferreira SH. Role of resi- blocked by 0.1% hydrogen peroxide in methanol for dent peritoneal cells in eosinophil migration induced by saline. 10 min after dehydrating the sections in ethanol. Slides Int Arch Allergy Immunol 1994; 104: 323–331. were rehydrated in ethanol and washed 2 × 5 min in PBS. 15 Lukacs NW, Strieter RM, Shaklee CL, Chensue SW, Kunkel SL. Biotinylated sheep anti-rat antibody (Amersham, Castle Macrophage inflammatory protein-1 alpha influences eosinophil recruitment in antigen-specific airway inflammation. Eur J Hill, NSW, Australia, 1:200 in PBS) was applied as a sec- Immunol 1995; 25: 245–251. ondary antibody. Following 30 min incubation and × 16 Stevceva L, Pavli P, Buffinton G, Wozniak A, Doe WF. Dextran 2 5 min washing in PBS the avidin-biotin complex sys- sodium sulphate-induced colitis activity varies with mouse tem (ABC kit, Vectastain, Vector Laboratories, Inc, Rock- strain but develops in lipopolysaccharide-unresponsive mice. dale, NSW, Australia) was used to detect the relevant J Gastroenterol Hepatol 1999; 14: 54–60.

Genes and Immunity Eosinophilia and IL-5 deficient mice L Stevceva et al 218 17 Saverymuttu SH, Camilleri M, Rees H, Lavender JP, Hodgson transport of serum-derived IgA into mucosal secretions. J Immu- HJF, Chadwick VS. Indium 111- scanning in the nol 1984; 132: 363–368. assessment of disease extent and disease activity in inflamma- 20 Husband AJ, Gowans JL. The origin and antigen-dependent dis- tory bowel disease. Gastroenterology 1986; 90: 1121–1128. tribution of IgA-containing cells in the intestine. J Exp Med 1978; 18 Sainte-Marie G. A paraffin embedding technique for studies 148: 1146–1160. employing immunofluorescence. J Histochem Cytochem 1962; 10: 21 Austyn JM, Gordon S. F 4/80, a monoclonal antibody directed 250–256. specifically against the mouse macrophage. Eur J Immunol 1981; 19 Sheldrake RF, Husband AJ, Watson DL, Cripps AW. Selective 11: 805–815.

Genes and Immunity