- Top of page
- MATERIALS AND METHODS
Background: In the adoptive transfer model of colitis it has been shown that regulatory T cells (Treg) can hinder disease development and cure already existing mild colitis. The mechanisms underlying this regulatory effect of CD4+CD25+ Tregs are not well understood.
Methods: To identify pathways of importance for immune regulation in protected mice we studied the genome-wide expression profile in the inflamed rectum of SCID mice with CD4+ T cell transfer colitis and in the uninflamed rectum of mice protected from colitis by Treg cells. We used DNA microarray technology (Affymetrix GeneChip Mouse Genome 430 2.0 Array), which enabled an analysis of a complete set of RNA transcript levels in each sample. Array results were confirmed by real-time reverse-transcriptase polymerase chain reaction (RT-PCR).
Results: Data were analyzed using combined projections to latent structures and functional annotation analysis. The colitic samples were clearly distinguishable from samples from normal mice by a vast number of inflammation- and growth factor-related transcripts. In contrast, the Treg-protected animals could not be distinguished from either the normal BALB/c mice or the normal SCID mice. mRNA expression profiles of cytokine, chemokine, and growth factor genes were significantly altered in colitic as opposed to noncolitic mice. In particular, the transcription factors STAT3, GATA2, and NFκB, the cytokine IL1β, and the chemokine receptors CXCR3 and CCR1 as well as their ligands all seemingly play central roles in the inflammatory processes.
Conclusions: We suggest that these molecules alone or in combination could be future therapeutic targets.
Crohn's disease (CD) and ulcerative colitis (UC), collectively termed inflammatory bowel disease (IBD), are chronic inflammatory disorders of the bowel. CD can affect the whole gastrointestinal tract, including the oral cavity and rectum, but in most cases it is localized to the distal small intestine; more than 1 part of the intestine can be affected at the same time. UC affects only the colon to various degrees; the initial site is the rectum. The etiology of IBD is unclear and a complex interplay between genetic and environmental factors is thought to contribute to the pathology. It is clear, however, that an excessive immune response to the commensal gut-associated bacteria is a hallmark of the intestinal inflammation associated with IBD, because animals raised in germ-free environments do not develop colitis in T-cell transfer models. When these animals are repopulated with normal gut-flora, intestinal inflammation occurs.1
Much information on the immune pathogenesis has been gained through preclinical animal models of colitis.2 One of these models, which histopathologically resembles human disease, is the aforementioned T-cell transfer model of colitis. The model was first described in 19933, 4; it was found that adoptive transfer of CD4+CD45RBHigh T cells into SCID or RAG2−/− recipients resulted in the development of a lethal disease characterized by diarrhea and wasting. More recently it has been shown also that unseparated CD4+ T cells, ConA-activated CD4+ T cells, CD4+CD25− T cells, and even CD4+CD45RBLow T cells can induce colitis in immunodeficient mice.5–8 The severity and kinetics of the disease development vary depending on the T-cell subset transferred. One advantage of the transfer model is that it is possible to distinguish between disease-inducing effector cells and regulatory cells that inhibit the disease. Thus, the model is well suited for studying mechanisms underling both the effector and the regulatory components of chronic inflammation. It was originally shown that cotransfer of CD4+CD45RBLow T cells together with the disease-inducing T cells was able to hinder disease development.9 Later, the protective capacity of these cells was found to be enriched among the CD25+ T cells.10–12 In agreement with these studies, we and others have shown that development of colitis induced by CD4+CD25− T cells can be prevented by cotransfer of CD4+CD25+ regulatory T cells (Treg) or unseparated CD4+ T cells derived from a 6-day coculture with immature dendritic cells.13 We have also shown that mice transplanted with CD4+CD25+ T cells alone do not develop colitis.14 Mottet et al15 reported that CD4+CD25+ Tregs are able to cure an already existing mild colitis in the adoptive transfer model of colitis. The mechanisms underlying the regulatory effect of CD25+ Tregs in colitis are not well understood. Although IL10 and TGFβ have been shown to play an important and complex role in the protective mechanisms of Tregs,10, 16–19 an unknown number of cytokines, chemokines, growth factors, and their receptors are most likely involved in both development and counterregulation of chronic colitis.
To obtain a full perspective on gene expression in the inflamed rectum and in the rectum of Treg-mediated protection, respectively, we studied the gene expression profile in rectum samples obtained from both colitic animals and animals protected from colitis by CD4+CD25+ Tregs. The primary aim was to identify pathways of importance for immune regulation in protected mice. We used DNA microarray technology (Affymetrix GeneChip Mouse Genome 430 2.0 Array, Santa Clara, CA) that enabled an analysis of a complete set of RNA transcript levels in each sample.
- Top of page
- MATERIALS AND METHODS
We analyzed the mRNA expression levels of rectum samples from mice with colitis and mice protected from colitis by Treg cells. The samples studied were chosen to obtain the highest possible degree of homogeneity. Thus, the colitic group of mice had a colitis score above 4 and both the protected and the control animals had a colitis score of 0. Mice with uniform levels of colitis were selected to minimize any inflammation-induced variations in the material. The small variation in levels of inflammation is also reflected in the uniform rectum weights of the colitic mice.
By using a PLS analysis the colitic samples were distinguished from the other samples by the first dimension. This dimension is predicted to be controlled in the negative direction of STAT3, GATA2, and NFκB signaling. We also searched the literature to confirm some predicted binding sites for NFκB, STAT3, and GATA2. On the list of promoters predicted to have NFκB binding sites, CCL536 and TLR237 had been shown to have functional NFκB binding sites. Likewise, the TNF1α promoter38 and the macrophage scavenger receptor 1 promoter39 were previously found to be regulated by STAT3, as predicted in our analysis. In general, however, only a tiny fraction of genes have had their promoter analyzed experimentally and no reports on promoter structure and function was found for most of the promoters predicted to contain potential binding sites for NFκB and STAT3, and this was also was the case for the 22 promoters predicted to have GATA2 binding. It is also important to note that the cis-element overrepresentation predictions are based on correlations between gene expression and predicted cis-elements in the promoters of the genes that dictate the negative direction of the first PLS dimension. The predictions are made in silico using a bioinformatic algorithm. Such correlations are useful to interpret the gene expression data and form hypotheses, but experiments are required to prove the functional importance of the predicted cis-elements for each individual promoter. Therefore, it is the identity of the cis-elements and the interacting transcription factor that is interesting, rather than the identity of the promoters predicted to contain the cis-elements. Along these lines it is therefore interesting that it has previously been shown that there is an increased activation of STAT3 in T cells isolated from the lamina propria from IBD patients. It is believed that STAT3 induces antiapoptotic factors, such as the BCL2 family, and thereby contributes to the accumulation of T cells in the lamina propria by hindering apoptosis.40, 41 Our results confirm the important role of STAT3 signaling in the inflamed colon and also indicate that STAT3 is 1 of the key proteins controlling the inflammatory response. NFκB has also been shown to play a major role in IBD, and established colitis was abrogated by administration of antisense oligonucleotide to the p65 subunit of NFκB in a murine model of IBD.42–44 It is interesting that GATA2 also controls the first PLS dimension; it has been shown that GATA2, together with GATA1, can modulate cytokine signals through protein-to-protein interactions, and inhibit STAT3, thereby regulating the growth and survival of hematopoietic cells.45 GATA2 has also been suggested to play a protective role in apoptosis. Since both STAT3 and GATA2 appear to control the PLS1 dimension, although they have opposite roles in apoptosis, this could indicate that a balance between these 2 signaling pathways is of importance for the inflammatory response in the colitic animals. To our knowledge, the importance of GATA2 in IBD has not been studied before, and it would be interesting to clarify its exact role. The control SCID and the protected group of mice are located slightly more negatively than the BALB/c control. This is in line with these animals having a low level of inflammation caused by the presence of Helicobacter infections in most animal facilities. The fact that the protected group does not differ from the control SCID group in localization on the PLS1 dimension shows that inflammation is not caused by the transplanted CD4+CD25+ Treg cells per se.
The second dimension identified by the PLS analysis distinguishes the SCID control and protected SCID samples from the other groups. This dimension is predicted to be controlled by the transcription factors NFY and E2F in the positive direction. E2F is a member of a family of proteins important in the control of transcription, cell cycle, and apoptotic processes in a very complex manner.46
As overfitting was a problem concerning the distinction between the SCID and protected groups, the model was not predictive in this regard and we could not predict any changes in mRNA expression in the rectum caused by the transplantation of Treg cells. This could be explained by the homogeneous nature of the samples, as all the mice in the protected group had a colitis score of 0, as did the control SCID. It has been shown that Treg cells transplanted simultaneously with disease-inducing cells mediate their function in the mesenteric lymph nodes and migrate only to the colon mucosa in a Treg-mediated therapeutic setting.47 This could explain why we were unable to detect any significant changes in the rectum samples of SCID mice cotransferred with Treg cells. Experiments are in progress to study mRNA expression in rectum samples from animals cured of mild colitis.
mRNA Expression Levels for Cytokine Genes
It is well documented that cytokines are important mediators of the intestinal inflammation associated with IBD. Many different cytokines have been implicated in the pathogenesis, including the proinflammatory and Th1 cytokines IL1, IL2, IL12, IFNγ, TNFα, IL18, and, more recently, TL1A and IL23, as well as the antiinflammatory and Th2 cytokines IL4, IL10, and IL13.48–51 In this study an overview of all transcripts is presented to identify new connections that might be of importance for IBD development. In particular, members of the cytokine IL1 superfamily have a higher expression level in the colitic animals. IL1β transcripts were more than 26-fold increased in the colitic animals compared to the other groups of mice. It has recently been shown that IL1β drives the proliferation and cytokine production by CD4+CD25+FoxP3− effector/memory T cells, but at the same time attenuates CD4+CD25+FoxP3+ Treg cell function, and allows escape of autoreactive CD4+CD25− effectors from suppression. In this way IL1β promotes autoantigen-specific effector cells, which in turn prevent Treg cells from maintaining tolerance to self-Ag.52 Our results suggest that these functions of IL1β are also of importance in the development of adoptive transfer colitis. Both IL1β and TNF, which we find significantly upregulated in the colitic group are known to be potent activators of NFκB.53
mRNA Expression Levels for Chemokine and Chemokine Receptor Genes
We have previously studied the expression of chemokine and chemokine receptor genes in mice protected from colitis by Treg cells.14 However, the study did not include the BALB/c and SCID mice as control groups. We therefore found it interesting to look at the expression levels of chemokine and receptor genes in this study too. The present results confirm and extend our previous results, showing that the expression of the chemokine receptors CXCR3 and CCR1 as well as their ligands CXCL9, CXCL10, CXCL11, CCL5, and CCL7 are expressed to a higher degree in the colitic animals than the protected animals. Therefore, these molecules might play a central role in controlling disease development. We have recently shown that transfer of CXCR3−/−CD4+CD25− T-cells resulted in an ameliorated form of colitis, with less severe clinical symptoms, which confirms the importance of this receptor in colitis development.54 Interestingly, there is a tendency that the protected group and the SCID control group have a higher expression level of these genes than the BALB/c group, albeit several-fold lower than the colitic animals. For instance, expression of CCL8 and CXCL9 was observed to be higher in control SCID mice compared to BALB/c mice. This indicates that these chemokines are involved in driving the low-level inflammation seen in SCID mice as a reflection of the presence of Helicobacter infections in standard animal facilities, including the facility at our institution.19 This low but significant level of inflammation can also be seen from the PLS analysis (Fig. 2) where the SCID samples are projected toward more negative values than the BALB/c samples along the first PLS dimension. The negative direction in this dimension is dominated by genes annotated with GO terms related to inflammation. Thus, these chemokines could be driving the low level of inflammation present in these animals due to inevitable Helicobacter colonization.
In particular, the expression of the chemokines CXCL2, CXCL5, and CXCL13 were found heavily upregulated (more than 30 times) in the colitic group compared to the other groups, which had very similar expression patterns. In our previous study we found transcripts for CXCL13 to be expressed at the same level in the colitic and in the protected group of mice and transcripts for CXCL5 to be below the detectable level.14 The samples included in our previous study represented a more diverse material (colitis score spanning from 2.5–4.5); therefore, we based our tables on median values from colitic animals with a colitis score of 3. In the present study, we selectively included animals with a colitis score of 4 or 5. The different findings could therefore be due to these chemokines playing a role in the late stages of disease with severe inflammation, rather than during early disease stages in less inflamed tissue. CXCL13 has previously been linked to the effector phase of murine experimental autoimmune encephalitis, where it plays a pathogenic role,55 an observation which is in line with our present data.
In our previous study14 we observed higher expression levels of the 2 inflammatory chemokines, CCL20 and CXCL16, in protected versus colitic animals, whereas the opposite pattern was observed in the present study. This discrepancy might also reflect the fact that the mice in the present study had more severe colitis than those in the former study. Thus, Treg activity at lower levels of inflammation might lead to a compensatory increase in the secretion of certain stromally derived chemokines (CCL20, CXCL16), whereas chemokine secreting cells might become insensitive to Treg activity at severe inflammation.
The expression patterns found in the array analysis was confirmed by real-time RT-PCR for IL1β, CCR5, IL2 receptor, γ chain, and TNFα.
mRNA Expression Levels for Genes Encoding Growth Factors and Their Receptors
A general observation in this study, related to the expression of the growth factor- (GF) and receptor-encoding genes, is that BALB/c mice and normal as well as Treg protected SCID mice express close to identical levels, whereas the colitic mice for some genes express highly increased and for other genes decreased levels of expression. The massive infiltration by monocytic cells and polymorphonuclear cells of the gut wall in colitic mice5 probably reflects the increased level of GM/M-CSF-receptor expression in the recipients (Table 3). In a previous study, we observed secretion of GM-CSF by splenic CD4+ T cells of mice with colitis, and 8–14 times higher frequencies of splenic granulocyte and macrophage colony-forming cells (CFC) in these mice compared with normal BALB/c mice.56 Thus, at the transcriptional level the scene is set for the observed influx of mature macrophages and granulocytes to the inflamed gut mucosa from both medullary and extramedullary sites. The down-modulation of fibroblast GF (FGF) 1 and FGF-receptor 2 and 3 expression in colitic mice appears somewhat odd, since all gut-wall tissues in transfer colitis undergo massive hyperplasia and hypertrophia.57 The attenuation of fibroblast GF-receptors in colitic mice might reflect that the inflammation-induced extensive hypertrophy of smooth muscular tissue occurs at the expense of the connective tissues. We find TGFβ1 RNA to be upregulated in colitic mice, although Treg cell-derived TGFβ has proven of major importance in controlling disease development in transfer colitis and even in curing the disease.18, 58 Apparently, the high level of TGFβ1 is insufficient to control the progressive course of the disease and probably reflects the massive influx of macrophages to the inflamed mucosa. Furthermore, several recent studies have clearly shown that TGFβ, in the presence of proinflammatory cytokines, e.g., IL6, induces the differentiation of the inflammatory T cells Th17.59–61 Thus, this highly pleiotropic cytokine can modulate the immune responses via multiple mechanisms dependent on the local environment. In active human IBD, increased mRNA levels of TGFβ have been associated with disease activity62; however, plasma levels of TGFβ1 seem not to be increased when compared to control persons.63 In another study including CD patients, the local protein levels of TGFβ1 were increased in biopsies from inflamed sites of the patients.64 Thus, although the serum level is not increased, it is possible that a local increase in TGFβ1 is associated with disease. It has been suggested that TGFβ1 may play a role in the repair processes of the gut wall in UC and thereby play a role in restoration of tolerance toward the bacterial flora.65–67
In conclusion, our results show that transcription factors STAT3, GATA2, and NFκB are involved in the inflammatory pathology in colitic mice, and specific inhibitors of these molecules might therefore dampen inflammation. Since the IL1 cytokine family, including IL1β, seems to play a central role in the inflammatory process, treatment with antagonistic antibodies to IL1β should be attempted to counteract inflammation. The chemokine receptors CXCR3 and CCR1 as well as their ligands CXCL9, CXCL10, CXCL11, CCL5, and CCL7 could also be new targets for antagonist treatment of patients with IBD.