Induction of inflammatory bowel (IBD)-like disease in mice by a targeted mutation in the Il10 gene (Il10−/−) is inbred strain dependent. C3H/HeJBir (C3) mice are colitis susceptible, whereas C57BL/6J (B6) mice are resistant. Genetic dissection of this susceptibility revealed 10 colitogenic quantitative trait loci (QTL). The aim of this study was to identify valuable candidate genes by a combination of QTL mapping and microarray analyses. Sixteen genes were differentially expressed between B6- and C3-Il10−/− mice and were located within the QTL intervals. Three major candidate genes (Cd14, Gbp1, Pla2g2a) showed prominent expression differences between B6- and C3-Il10−/− as well as between B6 and C3 wild-type mice, which was confirmed by semiquantitative or real-time RT-PCR. Because strain differences are known for Gbp1 and Pla2g2a, further analyses focused on Cd14. Western blot analysis revealed strain differences also on the protein level. Cd14 expression in animals with defective and intact Toll-like receptor (TLR)4 signaling (C3, C3H/HeN, B6, B6-Tlr4tm1Aki) make the TLR4 defect of C3 mice unlikely to be the reason for higher Cd14 expression. Less Cd14 expression in germ-free mice indicates a contribution of the microflora on Cd14 expression. Stimulation of naive peritoneal macrophages with bacterial antigens showed lower CD14 surface expression in B6 than in C3 mice. In conclusion, the large number of candidate genes was reduced to three major candidates that play an important role in inflammatory processes and immune response. Strain differences for them are already known or are shown in this study.
- gene expression
- inflammatory bowel disease
- quantitative trait locus
inflammatory bowel disease (IBD) in humans encompasses a heterogeneous collection of chronic inflammatory conditions, including Crohn's disease (CD) and ulcerative colitis. Development of IBD entails complex and as yet poorly understood interactions between the genetic and physical environments. The consensus opinion is that IBD results from the interaction of an environmental stimulus, such as antigens of the enteric bacterial flora, with host genetic factors that determine the individual's immune response or mucosal barrier function (43).
Animal models of chronic intestinal inflammation are valuable tools for the investigation of the genetic and environmental factors that appear crucial to the development of IBD (13, 21). These models enable investigators to analyze complex interactions among genes and between genes and the environment. One genetically engineered model with histological findings somewhat similar to those of human CD is the interleukin (IL)10-deficient (Il10−/−) mouse (29). IL10 is a cytokine that plays an important role in limiting the immune response against microbial antigens (4). In Il10−/− mice, the severity of the intestinal inflammation varies markedly as a function of the inbred strain background. For example, the C3H/HeJBir strain is highly susceptible to colitis, whereas the C57BL/6J strain develops only mild colitis in response to IL10 deficiency (10). This differential colitis susceptibility was genetically dissected in two previous linkage studies that reported 10 quantitative trait loci (QTL) associated with colitis susceptibility in segregating populations of Il10−/− mice (Cdcs1–10, cytokine deficiency-induced colitis susceptibility 1–10) (14, 35).
The ultimate goal of QTL mapping is to identify genes underlying a complex trait. However, identification of QTL genes is usually complicated by the large QTL intervals containing multiple candidate genes (16). The confidence intervals of the 10 colitogenic QTL identified in Il10−/− mice ranged from 10 to 40 cM (14, 35). The aim of this study was to identify by microarray analyses major candidate genes that are located within the intervals of the colitogenic QTL and differentially expressed in the colon of IBD-susceptible and -resistant strains before the onset of colitis.
MATERIALS AND METHODS
Conventional specified pathogen-free mice used for microarray experiments and for Cd14 real-time RT-PCR.
C3H/HeJBir.129P2 and C57BL/6J.129P2 mice congenic for the disrupted Il10 gene (formal designation Il10tm1Cgn) and control C3H/HeJBir.129P2 (C3) and C57BL/6J.129P2 (B6) strains homozygous for the wild-type (WT) Il10 allele were produced in a research vivarium at The Jackson Laboratory (TJL; Bar Harbor, ME) and then maintained by sibling matings at the Central Animal Facility, Hannover Medical School.
A total of 61 mice (29 ± 2 days of age; all mice weaned at the age of 21 ± 1 days) were produced and maintained in a room with a controlled environment (21 ± 2°C, 55 ± 5% relative humidity, 12:12-h light-dark cycle, 12–14 changes of air/h). Personnel entering the room were required to wear a gown, cap, surgical mask, overshoes, and gloves. Mice were housed separated by sex in individually ventilated cages (440-cm2 floor area) at a maximum of five animals on bedding of nonsterilized, dust-free, softwood fibers. Pelleted diet (Altromin 1314, Lage, Germany) containing 22.5% protein, 5.0% fat, and 4.5% fiber and tap water treated with UV light were provided ad libitum. Routine microbiological monitoring according to Federation of European Laboratory Animal Science Associations (FELASA) recommendations (40) did not reveal any evidence of infection with common murine pathogens except for Pasteurella pneumotropica. These mice are henceforth designated as conventional specified pathogen-free (conventional SPF) mice.
Five Toll-like receptor (TLR)4-deficient mice [B6.129P2-Tlr4tm1Aki (22), henceforth abbreviated Tlr4−/−] and five B6.129P2-Tlr4+/+ (Tlr4+/+) control mice were maintained in individually ventilated cages under similar conditions as for conventional SPF Il10−/− mice.
Germ-free mice used for Cd14 real-time RT-PCR.
A total of 30 mice (56 days of age) of both Il10−/− strains were taken from a colony maintained germ free (GF) in plastic film isolators (Metall+Plastik, Radolfzell-Stahringen, Germany). A pelleted irradiated diet (ssniff M-Z, Soest, Germany) containing 22.0% protein, 4.5% fat, and 3.9% fiber and autoclaved distilled water were provided ad libitum.
Rederived SPF mice used for Cd14 real-time RT-PCR, semiquantitative RT-PCR, fluorescence-activated cell sorting analysis, and Western blot analysis.
While this study was developing, we were able to rederive Il10−/− and WT mice into an SPF area protected by a more strict hygienic barrier system than the area of conventional SPF mice. These mice are henceforth designated as rederived SPF mice. C3H/HeN (C3/N) mice were maintained in the same area. Access to these mice was limited to few animal caretakers who had to pass through a water shower and were required to wear a gown, cap, surgical mask, overshoes, and gloves. Mice were housed separated by sex in open cages (360-cm2 floor area) at a maximum of five animals on bedding of autoclaved, dust-free, softwood fibers. Sterile pelleted diet (ssniff R-Z) and deionized, filtered, and UV light-treated water were provided ad libitum. Routine microbiological monitoring according to FELASA recommendations (40) did not reveal any evidence of infection with common murine pathogens. A total of 94 mice maintained under these conditions were used.
This study was conducted in accordance with German law for animal protection and with the European Communities Council Directive 86/609/EEC for the protection of animals used for experimental purposes. All experiments were approved by the Local Institutional Animal Care and Research Advisory Committee and permitted by the local government.
Two independent microarray experiments were performed, using conventional SPF mice: 19 for the first experiment (four C3 WT, five C3-Il10−/−, five B6 WT, five B6-Il10−/−) and 18 for the second (four C3 WT, four C3-Il10−/−, five B6 WT, five B6-Il10−/−). The mice were euthanized by cervical dislocation. Following a ventral midline incision, the colon (including the rectum but not the anus) was removed and flushed with PBS. Each colon was opened longitudinally with a pair of scissors, further dissected, and homogenized using a micropistill in 1.5 ml of peqGOLD Trifast FL (Peqlab, Erlangen, Germany). RNA was isolated according to the manufacturer's protocol, and quality was checked using RNA Nano LabChip technology (Agilent, Böblingen, Germany). For each of the two array experiments, equal amounts (7 μg) of RNA were pooled from the four or five animals of each mouse strain, resulting in a total of eight samples.
For biotin-labeled target synthesis starting from 3–5 μg of total RNA, reactions were performed using standard protocols supplied by the manufacturer (Affymetrix, Santa Clara, CA). Briefly, 3–5 μg of total RNA were converted to double-stranded DNA (dsDNA) using 100 pmol of a T7T23V primer (Eurogentec, Seraing, Belgium) containing a T7 promoter. The cDNA was then used directly in an in vitro transcription reaction in the presence of biotinylated nucleotides.
The concentration of biotin-labeled cRNA was determined by UV absorbance. In all cases, 12.5 μg of each biotinylated cRNA preparation were fragmented and placed in a hybridization cocktail containing four biotinylated hybridization controls (BioB, BioC, BioD, and Cre) as recommended by the manufacturer. Samples were hybridized to an identical lot of Affymetrix MG U74Av2 for 16 h.
Data selection and transformation procedures.
For normalization, all array experiments were scaled to a target intensity of 150, otherwise using the default values of the Microarray Suite (MAS) 5.0. Filtering of the results was done as follows. Genes were considered as strongly regulated if their fold change was greater than or equal to 2 or less than or equal to −2 and the statistical parameter for a significant change was less than 0.01 (change P value for changes called Increased) or greater than 0.99 (change P value for changes called Decreased). Additionally, the signal difference of a certain gene was greater than 200. Genes were considered as weakly regulated if their fold change was greater than or equal to 1.5 or less than or equal to −1.5, the statistical parameter for a significant change was less than 0.001 or greater than 0.999, and the signal difference of a certain gene was greater than 40. The data were submitted to the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO; GSM39288–GSM39297).
Selection of candidate genes.
Genes differentially expressed in C3- and B6-Il10−/− mice were assigned their chromosomal location with the help of databases [NetAffx (Affymetrix), Ensembl, UCSC, NCBI, TJL]. Differentially expressed genes that were found to be located within the colitogenic QTL intervals previously reported (14, 35) and that have a known function in immune and defense responses or that are thought to play a role in IBD in humans or animals were selected as candidate genes. Sequences that were used by Affymetrix to design templates of these genes were blasted [basic local alignment search tool; BLAST (2)] to verify the assigned gene names. Expression levels of these candidate genes were also compared in the WT mice to determine major candidate genes. Only genes whose expression differences were similar in WT mice to those detected in Il10−/− mice were considered to be major candidate genes, because we focused on differences due to the genetic background rather than due to the knockout of Il10.
A total of 48 mice (five conventional SPF and five GF mice of each Il10−/− strain, six rederived SPF mice of each WT and the C3/N strain, five B6-Tlr4−/− mice, and five B6-Tlr4+/+ mice) were used for real-time RT-PCR analysis. The proximal part of the colon (the area of macroscopically visible mucosal folds) was stored in RNAlater (Quiagen, Hilden, Germany) at −20°C until use for RNA extraction. RNA was extracted using the Nucleospin RNAII kit (Macherey Nagel, Düren, Germany) according to the manufacturer's protocol. From the extracted RNA, 1 μg was used for cDNA synthesis by Superscript II RT and oligo-dT and random hexamer primers (Invitrogen, Carlsbad, CA) according to the manufacturer's recommendations. Quantitative real-time RT-PCR was performed in an ABI PRISM cycler (Applied Biosystems, Foster City, CA), using a SYBR Green PCR kit from Stratagene (La Jolla, CA) and specific primers. A threshold was set in the linear part of the amplification curve, and the number of cycles needed to reach it were calculated. Relative mRNA levels were determined by use of a standard curve and by further normalization to the housekeeping gene Rps9. Melting curves were used to establish the purity of the amplified band. Primer sequences for Cd14 were 5′-CAT-TTG-CAT-CCT-CCT-GGT-TTC-TGA-3′ and 5′-GAG-TGA-GTT-TTC-CCC-TTC-CGT-GTG-3′, and those for Rps9 were 5′-CTG-GAC-GAG-GGC-AAG-ATG-AAG-C-3′ and 5′-TGA-CGT-TGG-CGG-ATG-AGC-ACA-3′. The annealing temperature was set to 55°C.
To detect differences in Cd14 mRNA levels between the groups of animals, one-way and two-way ANOVA and subsequent t-tests were performed, with the significance level adapted using a Bonferroni correction (P = 0.05 divided by the no. of comparisons).
Expression of Gbp1 and Pla2g2a was determined by semiquantitative RT-PCR in five B6 and five C3 WT mice (rederived SPF). The cDNA used as template was the same as for the real-time PCR of Cd14 in WT mice. Primer sequences for Gbp1 were 5′-AAG-AAG-TGA-AGC-AGG-GGA-CA-3′ and 5′-GCG-GCT-TCT-GCT-TTT-ATA-CG-3′. Primer sequences for Pla2g2a were 5′-ATG-GCC-TTT-GGC-TCA-ATA-CA-3′ and 5′-GGT-CTG-TGG-CAT-CCT-TGG-3′. Primer sequences for the housekeeping gene Rps9 are stated above. PCR was carried out using REDExtract-N-AMP-PCR ReadyMix (Sigma-Aldrich, Munich, Germany), annealing temperature was set to 55°C, and amplificates were loaded on a 3% NuSieve agarose gel (Biozym, Hamburg, Germany) containing SYBR Green (Gel Star, 4 μl/100 ml; Biozym, Hessisch Oldendorf, Germany) after various cycles of the PCR (up to 38 cycles were used).
Preparation of cecal bacterial antigens.
Cecal bacterial antigens were prepared by a method similar to that described by Cong et al. (11). Ceca were opened, placed in 1 ml of PBS, and vortexed. After removal of residual tissue and addition of DNase (10 μg/ml) to the remaining suspension, 1 ml of this bacterial suspension was added to 1 ml of glass beads. The cells were disrupted by vortexing for 3 min and then iced. The glass beads and unlysed cells were removed by centrifuging at 5,000 g for 5 min. The lysate was filter sterilized using a 0.45-μm syringe filter. Sterility of the lysate was proofed by culturing an aliquot in thioglycollate broth (Oxoid, Basingstoke, UK) at 37°C for 1 wk.
Culture of peritoneal macrophages and fluorescence-activated cell sorting analysis.
Macrophages from 15 C3-Il10−/− and from 9 B6-Il10−/− rederived SPF mice were harvested by peritoneal lavage with 10 ml of ice-cold culture medium (RPMI 1640 with 1% l-glutamine; Biochrom, Berlin, Germany), 10% heat-inactivated fetal calf serum (Biochrom, Berlin, Germany), 100 U/ml penicillin G and 100 μg/ml streptomycin (GIBCO Invitrogen, Karlsruhe, Germany), and 100 μM β-mercaptoethanol (GIBCO Invitrogen). For each experiment, macrophages of three mice were pooled, washed twice by centrifuging at 500 g, and plated on three polystyrene cell culture dishes (Greiner Bio One, Frickenhausen, Germany) at a density of 1 × 106 in 1 ml of culture medium at 37°C and 5% CO2. After 2 h of cultivation, medium was removed, and the cells were washed twice to remove nonadherent cells. Hereafter, macrophages were grown for 44 h either in pure medium, medium containing interferon (IFN)γ (10 U/ml; Relia Tech, Braunschweig, Germany), or medium containing IFNγ and cecal bacterial antigens (20 μl lysate/ml) before they were collected by manual scraping in culture medium. Cells were stained with a FITC-conjugated rat anti-mouse CD14 antibody (BD Biosciences, Heidelberg, Germany).
Western blot analysis.
A total of 16 rederived SPF mice (10 ± 1 wk of age), 4 of each Il10−/− and WT strain, were used for studying CD14 protein expression in the colon and in the ileum. Whole cell lysates from colonic and ileal mucosa (obtained by manual scraping) of two mice of each strain were pooled and subjected to SDS-PAGE, followed by blotting on a nylon membrane and incubation overnight with the CD14 antibody mentioned above and with a rabbit anti-mouse β-actin antibody (Sigma-Aldrich, Munich, Germany). Peroxidase-conjugated goat anti-rat IgG for anti-CD14 and peroxidase-conjugated goat anti-rabbit IgG for anti-β-actin were used as secondary antibodies (Dianova, Hamburg, Germany). A lysate of CD14-positive macrophages (cell line J774) served as positive control. Western blots were replicated once.
Seven 4-wk-old B6-Il10−/− and C3-Il10−/− conventional SPF mice (littermates of the animals taken for microarray analyses), six 15- to 18-wk-old B6-Il10−/− mice, and fourteen 15- to 20-wk-old C3-Il10−/− GF mice as well as twenty-seven B6-Il10−/− and nine C3-Il10−/− rederived SPF mice (up to 40 wk of age) were used for histological examination of the cecum and colon as described previously (8). The mice were euthanized by CO2 asphyxiation. The colon was prepared as a “Swiss Roll” (37) without being opened before rolling, processed routinely for histological examination, and stained with hematoxylin and eosin. Histopathological lesions were graded separately for the proximal, middle, and distal colon from 0 (= no inflammation) to 3 (= severe inflammation) using a scoring system that focused on cell infiltration, ulceration, and hyperplasia (8).
Determination of inflammatory changes in the colon of Il10−/− mice.
Colons of littermates of the animals taken for microarray analyses (4-wk-old Il10−/− conventional SPF mice) were examined for histopathological changes. No inflammatory lesions were observed in the colon of B6-Il10−/− animals. In C3-Il10−/− mice, two littermates of animals taken for the first array experiment showed mild inflammatory lesions in the colon (mean score of 0.7 and 1.3, respectively, in the three colonic segments evaluated in these 2 animals). No lesions were detected in the colon of any other C3-Il10−/− SPF mouse examined histologically. These conventional SPF mice were taken for microarray analyses, because they were maintained in the same environment as the mice that were taken for previous mapping studies. Confirmation of Cd14 expression differences in Il10−/− mice was also performed using these mice.
Consistent with previous studies (47), none of the 20 Il10−/− GF mice (up to 20 wk of age) showed signs of intestinal inflammation. No colonic lesions were seen in the 36 rederived SPF mice (up to 40 wk old). Because these mice did not develop colitis up to the age investigated, we decided to use them and their WT control mice for further experiments. Hereby the possibility was excluded that colitis development affected the results.
Comparing gene expression in Il10−/− mice: eight candidate genes.
By comparing gene expression in colonic tissue of both Il10−/− strains, we identified 94 genes (106 probe sets, some genes represented by >1 probe set) that showed constantly different expression levels in both array experiments (Table 1). Of these genes, 16 were located within the candidate gene regions (Table 1, underlined). Eight of these genes were selected as candidates because of their known function in immune response (Fig. 1). Genes that were not considered candidates on the basis of their function are located on chromosome (Chr) 2 (Capn3 at 67.2 cM), Chr 3 (Adh1 at 71.2 cM, Bglap1 and Bglap2 at 42.6 cM), Chr 4 [1300002F13Rik (MIG6 homolog) at ∼79 cM, Tceb3 at ∼65 cM], Chr 8 (Mtmr7 at ∼22.5 cM), and Chr 17 (Glo1 at 29.8 Mb). The use of different databases did not permit incontrovertible assignment of the genes represented by the probe sets 101787_f_at, 97181_f_at, 98254_f_at, and 93909_f_at. These probe sets very likely represent intracisternal A particles (Iap), and at least one of them (Iap GenBank ID M10062) could be located within the Cdcs5 interval on Chr 17.
Comparing the expression of candidate genes in WT mice: three major candidate genes.
By also comparing gene expression of the candidate genes in WT animals, we discovered that three genes [Cd14; guanylate nucleotide-binding protein-1 (Gbp1); and phospholipase A2, group IIA (Pla2g2a)] showed a major contribution of the genetic background on the expression level, because the expression differences identified between both Il10−/− strains were also detected in both WT strains (Figs. 2 and 3). Furthermore, these genes showed very distinct expression differences between the C3 and B6 genetic backgrounds. Gbp1 (26) and Pla2g2a (48) are already known to be polymorphic between C3 and B6 mice, and a polymorphism in the human CD14 gene is associated with IBD in humans (27, 41). Therefore, these genes were considered to be the major candidate genes.
Because the Cd14 gene product is a monocyte/macrophage marker, expression data were compared for other surface markers of these cells [Itgam (CD11b), Fcgr3/2/1 (CD16/32/CD64), Cd68] in Il10−/− mice. None of these markers was differentially regulated between C3- and B6-Il10−/− mice (according to the filter criteria mentioned above).
Confirming expression differences of Gbp1, Pla2g2a, and Cd14.
Expression differences of Gbp1 and Pla2g2a were confirmed by semiquantitative RT-PCR in colonic tissue of 8-wk-old B6 WT and C3 WT mice. Levels of mRNA of both genes were considerably higher in all C3 mice than in B6 mice (Fig. 4A). Microarray data for Cd14 were confirmed by real-time RT-PCR using colonic tissue from conventional SPF Il10−/− mice (Fig. 4B; B6-Il10−/− SPF mice compared with C3-Il10−/− SPF mice, P = 0.0008) and WT mice (Fig. 4C; B6 compared with C3 mice, P = 0.0098). Furthermore, expression of Cd14 was measured in B6- and C3-Il10−/− mice maintained under GF conditions, because GF Il10−/− mice did not develop IBD and because of the absence of antigens from living microorganisms in GF intestines. There was a markedly higher expression of Cd14 in C3-Il10−/− GF mice than in B6-Il10−/− mice maintained in either a GF or conventional SPF environment (Fig. 4; C3-Il10−/− GF compared with B6-Il10−/− GF and compared with B6-Il10−/− SPF, P < 0.0001 in both cases). Two-way ANOVA revealed significantly higher Cd14 mRNA levels in conventional SPF than in GF mice (P = 0.0024), indicating that bacterial flora increases Cd14 expression.
Because differences in the function of Pla2g2a and Gbp1 are already known between B6 and C3 mice, only Cd14 was selected for further analyses.
Defective TLR4 signaling of C3 mice is unlikely to be the reason for the strain differences identified.
Because C3 mice are LPS hyporesponsive due to a defect in TLR4, the question arose as to whether higher Cd14 expression in SPF mice compensates for this defect. Therefore, expression of Cd14 was compared between B6 WT, C3 WT, and C3/N mice, the last of which expresses a functional TLR4, as well as between B6-Tlr4−/− and B6-Tlr4+/+ mice. As in TLR4-deficient C3 mice, a significantly higher Cd14 expression was detected in LPS normal-responsive C3/N mice than in B6 WT mice (P = 0.0004). Furthermore, C3/N mice express more Cd14 mRNA than C3 WT mice (Fig. 4C; C3/N compared with C3, P = 0.0156). No difference in colonic Cd14 expression was observed between B6-Tlr4−/− and B6-Tlr4+/+ mice (Fig. 4D). This indicates that the TLR4 deficiency of C3 mice is unlikely to be the reason for higher Cd14 expression in these mice compared with B6 mice.
Strain differences in the Cd14 mRNA level correlate with differences in the protein level.
CD14 protein was clearly detectable in colonic mucosa of C3 WT and C3-Il10−/− mice by Western blot analysis. In the B6 background, there was only slight expression of CD14 in the colon of both WT and Il10−/− mice; similar results were obtained in the ileum of both strains (Fig. 5). Thus the differences between B6 and C3 in CD14 protein expression are consistent with differences identified on the mRNA level.
In vitro Cd14 response to environmental stimuli is markedly different between B6- and C3-Il10−/− mice.
The immune system of the mucosa has to deal with different kinds of bacterial components, IBD in Il10−/− mice is dependent on the bacterial environment, and real-time RT-PCR revealed a higher colonic Cd14 expression in SPF mice than in GF mice (Fig. 4B). Therefore, we tested whether the Cd14 response to antigens of intestinal bacteria is strain dependent. Naive peritoneal macrophages were obtained from C3-Il10−/− and B6-Il10−/− mice, cultured for 44 h either in the presence of cecal bacterial antigens or left without stimulation, stained with a CD14 antibody, and analyzed by fluorescence-activated cell sorting (FACS). As seen in Fig. 6A, stimulated macrophages from C3-Il10−/− mice showed a higher expression of the CD14 protein than macrophages from B6-Il10−/− mice. Furthermore, there is a clear increase in CD14 expression on macrophages from C3-Il10−/− mice after stimulation compared with unstimulated macrophages (Fig. 6B), while macrophages from B6-Il10−/− mice show almost no response to cecal bacterial antigen stimulation (Fig. 6C).
Using a combination of QTL mapping and microarray analyses, we were able to identify eight candidate genes that may contribute to differential colitis susceptibility in B6- and C3-Il10−/− mice. These genes are located within one of the candidate gene intervals determined in previous studies by QTL mapping (14, 35) and show expression differences between colitis-susceptible C3-Il10−/− mice and colitis-resistant B6-Il10−/− mice. Among these genes, Cd14, Gbp1, and Pla2g2a were considered to be major candidates, because the expression differences identified between both Il10−/− strains were also detected in both WT strains, and these genes showed the most distinct expression differences between the C3 and B6 genetic backgrounds.
This study was designed to detect major expression differences in total colonic tissue between both genetic backgrounds. One positive effect of using total colon is the very short time of 30 s maximum necessary for handling the tissue before it is preserved in an appropriate buffer. This very likely allowed us to mirror the in vivo gene expression. The age of the animals taken from an SPF environment was set to 4 wk, because colitis usually does not occur in animals of this age in our facility. Nevertheless, histological examination revealed a mild inflammatory response in C3-Il10−/− animals used for the first experiment but not in those used for the second. This is very likely the reason why some of the candidate genes showed a more prominent expression difference in the first than in the second array experiment. However, expression differences for the major candidate genes were detectable also in WT mice and were confirmed using RT-PCR approaches.
Two previous studies identified 10 colitogenic QTL (14, 35). A major QTL (Cdcs1) was detected in a linkage study using an F2 population derived from strains C3- and B6-Il10−/− (peak at 61.8 cM) and replicated using a backcross population to C3-Il10−/− (peak at 68 cM). With the use of an expanded backcross population, two peaks were identified in the Cdsc1 interval at 60.5 and 70.3 cM on Chr 3 (8). Multiple peaks in the Cdsc1 interval might indicate that several genes contribute to this linkage and that different genes contribute to this QTL to a different extent in the different studies (34). This is underlined by the fact that the linkage study using the F2 population and the studies using the backcross populations were conducted in different environmental contexts, and the Il10−/− mouse model is characterized by complex intergenetic and gene environmental interactions (7, 34).
Within this main candidate gene interval, one of the major candidate genes, Gbp1, is located at 67.4 cM. Members of the GBP family of GTPases such as GBP1 are among the most abundant proteins induced by IFNγ via the transcription factors signal transducer and activator of transcription 1 (STAT1) and IFN regulatory factor 1 (IRF1) (9, 44). IFNα/β, TNFα, IL1α, and IL1β also induce expression of these proteins (32). GBP1 has anti-viral activity and mediates the anti-proliferative effect of proinflammatory cytokines (3, 17). Interestingly, some mouse strains, including B6 mice, are refractory to the induction of Gbp1 mRNA, whereas C3 mice are not (48). This is consistent with our microarray data. The human homolog is located on Chr 1p, within a region that showed association to Crohn's disease in one study, but this association was not replicated (23). The susceptibility locus for human IBD on Chr 1p, called IBD7, is located further distal on this chromosome.
Pla2g2a, for which a polymorphism is known between C3 and B6 mice, is located in the interval of Cdcs9 on Chr 4 (Fig. 1). B6 mice are homozygous for a defective Pla2g2a allele bearing a frameshift null mutation (26), also known as Mom1 (modifier of Min, multiple intestinal neoplasia) (33, 38). The Pla2g2a gene product has antimicrobial activity on mucosal surfaces (20). Furthermore, the Pla2g2a gene product is a key enzyme in eicosanoid synthesis and is therefore an interesting candidate gene in the context of inflammatory lesions (6). It has been suggested that this enzyme plays an essential role in human IBD, because the activity of its synthesis was increased in patients with ulcerative colitis and Crohn's disease (18, 19). In the Il10−/− model, the C3 allele of the corresponding QTL contributed to colitis susceptibility. Thus high expression of Pla2g2a in colitis-susceptible C3 mice might contribute to IBD susceptibility because of harmful effects of the enzyme products on the mucosa. This is corroborated by a recent study, in which experimental IBD was ameliorated by an extracellular phospholipase A2 inhibitor (28). The human homolog is located near a susceptibility locus for human IBD (IBD7).
The murine Cd14 gene is located within the Cdcs6 interval on Chr 18 (Fig. 1) and the human homolog within a susceptibility region for human IBD on Chr 5q (IBD5). The gene product is a receptor constitutively expressed on the surface of monocytes, macrophages, and neutrophils and is also present in a soluble form in serum (31). It is essential for LPS-dependent signal transduction via TLR4 (54). It also recognizes other bacterial cell wall-derived components, such as peptidoglycan (PGN), and interacts with TLR2 (39). A polymorphism in the human CD14 promoter (T/C at position −159) is related to high expression of CD14 (5). This polymorphism was shown to be associated with Crohn's disease and ulcerative colitis (27, 41).
C3H/HeJ mice are LPS hyporesponsive due to a defect in TLR4. Therefore, the question arose as to whether higher Cd14 expression in this strain might have been a compensatory effect. However, expression differences were confirmed in GF animals without stimulation from living bacteria, and analysis of mice with defective and functional TLR4 signaling showed higher Cd14 mRNA levels in C3/N than in LPS hyporesponsive C3 mice. Furthermore, a targeted mutation of TLR4 did not change Cd14 expression in B6 mice. This indicates that the TLR4 defect of C3 mice is not the reason for higher Cd14 expression in this strain.
Also, confirmation of Cd14 expression data using GF Il10−/− mice that did not develop IBD made inflammatory changes in the colon of C3 animals, e.g., a higher density of macrophages, unlikely to be the reason for a higher Cd14 expression in the C3 background. This is corroborated by the data from the microarray experiments using conventional SPF mice, which showed no upregulation of other surface markers on macrophages in C3-Il10−/− mice compared with B6-Il10−/− mice. Furthermore, Cd14 was also upregulated in C3 WT mice, which did not develop IBD. Notably, IBD has been described in C3H/HeJBir mice (49) (which served as a control in this study), but in our animal facility, these mice have never shown inflammatory changes in the intestine.
The identified strain differences between B6 and C3 in colonic Cd14 mRNA expression were also detected at the protein level (Fig. 5), and this observation correlated with results of in vitro experiments in which C3 macrophages responded with higher CD14 expression than B6 macrophages on stimulation with bacterial antigens. Furthermore, CD14 expression showed a distinct difference between unstimulated and stimulated C3 macrophages, whereas this was not the case for CD14 expression on B6 macrophages (Fig. 6). Bacterial stimuli are necessary for the development of colitis in a variety of mouse models, including the Il10−/− mouse, and they also play an important role in the pathogenesis of human IBD (7, 11, 43). We further observed that Cd14 expression is higher in SPF than in GF mice. Thus differential CD14 expression might provide a link between IBD susceptibility and the host response to intestinal bacteria.
A possible role of CD14 in gut homeostasis is further corroborated by studies in which CD14 has been identified in the colon of mice (36, 42) and rats (50) and in which increased colonic expression of CD14 was demonstrated during the course of experimental colitis (36, 42). However, blockade of CD14 with antibodies during experimental Shigella flexneri infection resulted in a 50-fold increase in bacterial invasion and more severe tissue injury of the intestinal mucosa (52, 53). The latter study (with low CD14 levels contributing to tissue injury) fits to the linkage study performed by Farmer et al. (14), in which the B6 allele of the QTL containing Cd14 contributed to colitis susceptibility. This corresponds to a low expression of Cd14 in the large intestine. A key could be the interaction of CD14 with different TLRs, the signaling of which results in specific immunoregulatory cytokine release. For example, TLR2-mediated signals seem mainly to induce a Th2 cytokine release, whereas TLR4 activation mainly leads to a Th1 response (45, 46). In addition to membrane-bound CD14, the soluble form of this protein is discussed to play a role in maintenance of gut homeostasis (15, 30, 51).
Hundreds of interesting candidate genes are located within the confidence intervals of the 10 colitogenic QTL identified in Il10−/− mice, making the identification of QTL genes challenging. Combination of gene expression analysis and QTL mapping has been successfully applied to detect major candidate genes or QTL genes in a variety of models of complex diseases (1, 12, 24, 25). With the approach used in this study, the number of genes that should be examined preferentially for their role as modifier genes for experimental IBD was reduced to eight genes. For the three major candidate genes, strain differences are either already known or have been demonstrated in this study.
This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB-280, H. J. Hedrich and M. Mähler; GK-705, A. Bleich) and from the Medizinische Hochschule Hannover (HILF I, A. Bleich).
We gratefully acknowledge S. Akira for providing Tlr4−/− mice; E. Leiter for remarks on the manuscript; and I. Köhn, A. Smoczek, T. Töpfer, and R. Wehling for excellent technical assistance.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: A. Bleich, Central Animal Facility and Institute for Laboratory Animal Science, Hannover Medical School, Carl-Neuberg-Str. 1, 30625 Hannover, Germany (e-mail:).
- Copyright © 2006 the American Physiological Society