DX5+NKT cells induce the death of colitis-associated cells: involvement of programmed death ligand-1



NKT cells are activated by CD1d and show an immune regulating function. Here, we investigated whether DX5+NKT cells could be used to reduce colitis in a chronic colitis mouse model and studied the potential immunological mechanisms involved. Chronic colitis was induced either by transfer of enriched CD62L+CD4+ T cells to severe-combined-immunodeficient mice or by feeding dextran sodium sulfate to immune competent mice. DX5+NKT cells were transferred to mice with chronic colitis. Co-transfer of DX5+NKT cells, but not CD8+ control cells, prevented the onset of colitis, and the immune regulatory effect of DX5+NKT cells was completely abrogated by injecting CD1d blocking antibody. Moreover, DX5+NKT cells reduced established colitis in both chronic colitis models. In vitro, DX5+NKT cells induced cell death of colon-infiltrating lymphocytes isolated from diseased mice. This effect was inhibited in the presence of either anti-CD1d or anti-programmed death ligand-1 (PD-L1) blocking antibodies. The specific potency of DX5+NKT cells in regulating chronic colitis in two mouse models is demonstrated. In vitro testing suggests that DX5+NKT cells activated by CD1d induce cell death of colitis-inducing lymphocytes, which is mediated through PD-L1. Therefore, DX5+NKT cells could be important in the regulation of immune responses associated with chronic colitis.


dextran sodium sulfate


magnetic activated cell sorting


programmed death ligand-1




Despite intensive research the etiology and pathogenesis of inflammatory bowel disease remains unclear. Many studies have focused on the pivotal role of T cells in the development of colitis. In this respect, Powrie et al. 1 have shown that the CD45RBhigh fraction of splenic CD4+ cells isolated from healthy mice causes colitis in severe-combined-immunodeficient (SCID) mice within 12 weeks after transfer. Transferred CD45RBhigh cells migrate to the mucosa of the colon and exhibit a Th1-like cytokine production pattern, with increased levels of IFN-γ, TNF-α, and IL-2 2. Non-fractionated CD4+ cells, including CD45RBhigh and CD45RBlow cells, are also able to induce colitis in SCID mice, albeit at a slower 24-week pace 3. Moreover, it is also known that a normal intestinal flora is necessary to induce colitis, because SCID mice reared under germ-free conditions do not develop colitis after cell transfer 4. Together, these studies suggest that naïve T cell activities left unchecked by balancing regulatory cell populations in the presence of bacterial antigen promote an uncontrolled immune response, leading to colitis. In the present study we planned to use a similar colitis model system in mice to investigate a potential immune modulating function of NKT cells.

The NKT cell has recently emerged as a cell with regulatory potential relevant for various diseases of the immune system. In general, NKT cells co-express an NK cell marker and T cell receptors with an invariant α and a polymorphic β chain. In mice these cells are CD4+CD8 or CD4CD8, and express an invariant T cell receptor α chain encoded by the Vα14 and Jα281 (now known as Jα18) gene segments 5, 6. In humans, NKT cells express the Vα24 and Jα18 gene segments preferentially associated with Vβ11 6, 7. Another important feature of NKT cells is that they are activated via CD1d, which belongs to a group of MHC class I-like molecules (50 kDa) called CD1 8. CD1 molecules are associated on the surface with β2-microglobulin 9, and are known to present lipids including glycosylceramides and glycosylphosphatidylinositol 10, 11. In humans there are two groups of CD1 molecules (CD1a, b, c, e and CD1d), but only the second group has been described in mice (CD1d.1 and CD1d.2) 12, 13. CD1d is expressed on hematopoietic cells 14 and on epithelial cells of the intestine 15, where it is thought to stimulate NKT cell production of IL-4 and IFN-γ 16, and to increase their cytolytic activity 17. Therefore, studies have suggested a pivotal role of NKT cells in regulation of the immune response. For example, activation of NKT cells by α-galactosylceramide provides protection not only against dextran sodium sulfate (DSS)-induced acute colitis 18, but also against experimentally-induced autoimmune encephalomyelitis (EAE) 19, 20. Effects of NKT cells on the development of autoimmune Type 1 diabetes in animals 21, 22 and on long-term survival of corneal allografts have also been reported 23. In contrast, IL-13-producing NKT cells mediate oxazolone colitis in a Th2 colitis model 24. However, to date, little is known about how NKT cells fulfill their immune-regulating function.

In the present study we investigated the potential of DX5+NKT cells to suppress colitis development and to reduce colitis after onset 25. For this purpose, CD62L+CD4+ T cells, which express high levels of CD45RB 26, 27, were used to induce colitis in mice 28. We show that DX5+NKT cells can indeed inhibit colitis development, and furthermore, can reduce colitis after it has been established. Moreover, we demonstrate that DX5+NKT cells also improve chronic colitis in the DSS model, whereby this sulfated polysaccharide reproducibly induces chronic colitis in immune competent BALB/c mice 29. Mechanistically, we show that CD1d is up-regulated on colonic epithelial cells, and that DX5+NKT cells require interaction with CD1d to mediate their immune regulatory effects against colitis. In addition, we demonstrate in vitro that DX5+NKT cells induce the death of CD62L+CD4+ T cells and DSS colitis cells after interaction with CD1d. Importantly, this killing effect is mediated in part via programmed death ligand-1 (PD-L1; also known as B7-H1), a member of the B7-CD28-family molecules linked to cell death.


Transfer of CD62L+CD4+ T cells induces severe colitis in BALB/c SCID mice

CD62L+CD4+, CD62LlowCD4+ and DX5+NKT cell populations were isolated from the spleens of BALB/c mice by MACS. Flow cytometry analysis of the separated cell populations using a antibody panel (see Section Materials and methods) revealed a high purity of the isolated populations, as shown in Fig. 1. To determine the effect of different T cell populations on the development of colitis, SCID mice first were injected with either 1 × 106 CD62L+CD4+, or CD62LlowCD4+, T cells. Approximately 6 weeks after cell injection, mice receiving the CD62L+CD4+ T cells started to show weight loss (Fig. 2A), with two of the six mice developing severe outward signs of colitis, including a prolapsed rectum 30. In contrast, mice receiving CD62LlowCD4+ cells showed no clinical signs of colitis, and continued to gain weight (data not shown). After 10 weeks, mice were killed and distal colon tissue was collected for histological analysis. Histological evaluation of the colon confirmed severe colitis in mice receiving CD62L+CD4+ T cells, with evidence of epithelial damage, loss of crypts and inflammatory infiltrates (Fig. 2B,C).

Figure 1.

FACS analysis of CD62L+CD4+, CD62LlowCD4+ and DX5+NKT cells after isolation from the spleen of BALB/c mice. (A) CD62L+CD4+ T cells were isolated by MACS and analyzed with FITC-conjugated anti-mouse-CD4 and PE-conjugated anti-mouse-CD62L antibody. (B) CD62LlowCD4+ cells were collected in the negative fraction. (C) After isolation of DX5+NKT cells by MACS the cells were labeled with FITC-conjugated anti-mouse CD3 molecular complex and PE-conjugated anti-mouse CD49b (clone: DX5) antibody. The isolation of DX5+NKT cells was performed multiple times (n=15) and the purity ranged from 86% to 96%, with an average purity of 91%. The dot-plots on the left are for appropriately matched isotype control antibodies in each case.

Figure 2.

Mouse weight measurements were regularly made after CD62L+CD4+ T cell transfer to BALB/c SCID mice. Experimental groups consisted of mice receiving no further treatment, or treatment with either DX5+NKT cells, or CD8+ T cells (control cells). Results from these experiments are shown as the median value (% of starting weight) from six mice in each experimental group. (A) No weight loss was detected after co-transfer of CD62L+CD4+ and DX5+NKT cells. Control mice injected with only CD62L+CD4+ T cells, or CD62L+CD4+ and CD8+ T cells showed weight loss after approximately 6 weeks. (B) Histological specimens were obtained from BALB/c SCID mice of each group 10 weeks after transfer of CD62L+CD4+ T cells. Photomicrographs of hematoxylin and eosin-stained distal colon sections show: epithelial damage, loss of crypts, and a severe inflammatory infiltrate in a mouse injected with CD62L+CD4+ T cells, but relatively normal colon tissue from a mouse treated with DX5+NKT cells. Treatment with CD8+ T cells did not reduce the colitis. (C) Using a histological scoring system of 0–4 (4 being the most severe), all mice from each group were evaluated. The mean ± SEM of the score is shown. *represents p<0.05 between the indicated groups.

DX5+NKT cells prevent colitis induced by CD62L+CD4+ T cells

Next we tested whether DX5+NKT cell therapy could affect the development of colitis in this model. DX5+NKT cells were isolated from the spleens of BALB/c mice by MACS using DX5 and CD3 antibodies (Fig. 1). Purified DX5+NKT cells were injected into SCID mice shortly before (day –2), and periodically after (2, 4 and 6 weeks), colitis induction with CD62L+CD4+ cell injection. In contrast to mice receiving only CD62L+CD4+ T cells, those mice also receiving DX5+NKT cells showed neither weight loss (Fig. 2A), nor any outward signs of colitis. The degree of colitis was confirmed histologically by scoring of the colonic tissues after 10 weeks. Indeed, the analysis revealed only mild colitis after co-transfer of CD62L+CD4+ and DX5+NKT cells, contrasting with evidence of severe colitis in the CD62L+CD4+ control group (Fig. 2B, C). As a further control for the DX5+NKT cell effect, we also co-transferred CD8+ T cells with colitis-inducing CD62L+CD4+ T cells. Results from this experimental group showed mice developing colitis with clinical signs of disease (weight loss, anal prolapse) (Fig. 2A). Histological analysis of mice killed 10 weeks after co-transfer of CD62L+CD4+ and CD8+ cells confirmed the development of severe colitis (Fig. 2B, C). These results suggest that DX5+NKT cells can specifically prevent colitis development in a murine CD62L+CD4+ cell-transfer model.

CD1d expression is up-regulated with colitis and is necessary for the protective effect of DX5+NKT cells

It is known that NKT cells are activated by the MHC class I-like molecule CD1d, which is expressed on hematopoietic cells and intestinal epithelium. One of our objectives was to determine if CD1d expression is increased in colonic epithelium of mice with colitis, which could provide a potential mechanism by which NKT cells might be stimulated to affect the progression of this disease. Under normal conditions our immunohistochemical results show that CD1d is expressed on the surface of colon epithelial cells (Fig. 3A). Importantly, after induction of colitis by CD62L+CD4+ T cell transfer, the immunohistochemical analysis showed increased expression of CD1d on epithelial cells (Fig. 3A). To support this finding, we further examined CD1d mRNA expression in colonic epithelial cells. Using real-time RT-PCR our results show that CD1d mRNA was increased approximately twofold in mice with colitis after receiving CD62L+CD4+ T cells, versus healthy mice and mice receiving CD62LlowCD4+ T cells (Fig. 3B). Therefore, induction of chronic colitis by CD62L+CD4+ T cells in BALB/c SCID mice up-regulates CD1d expression in epithelial cells of the colon.

Figure 3.

The potential role of CD1d in the DX5+NKT cell effect was assessed. (A) CD1d expression was tested by immunohistochemistry on epithelial cells of the colon of healthy BALB/c SCID mice, and mice injected with CD62L+CD4+ or CD62LlowCD4+ T cells. CD1d expression was increased on epithelial cells of CD62L+CD4+ T cell-injected mice 10 weeks after induction of colitis, versus healthy or CD62LlowCD4+ T cell-injected mice. (B) cDNA was obtained from colonic epithelial cells isolated from these same animals. Real-time RT-PCR was performed and quantification of CD1d was estimated using standard curves. CD1d was up-regulated after transfer of CD62L+CD4+, but not CD62LlowCD4+ T cells (versus healthy mice). The experiment was repeated twice (n=6/group for each experiment) and the data represent the mean ± SEM. (C) In vivo, anti-CD1d blocking antibody administered with the usual CD62L+CD4+ induction, and DX5+NKT cell treatment, abrogated the protective effect of DX5+NKT cells. Once again, the experiment was repeated twice (n=6/group for each experiment) and the data represent the mean ± SEM. (D) Histological evaluation of colon specimens from these mice show that treatment with DX5+NKT cells and anti-CD1d antibody results in severe colitis, while the use of isotype control antibody does not interfere with the protective effects of DX5+NKT cells. (E) Histological scoring of the tissues from these same mice based on the above-mentioned scoring system is shown (mean ± SEM). p<0.05 when two corresponding groups share matching symbols.

To determine if CD1d molecules were necessary for mediation of the protective effect of DX5+NKT cells in chronic colitis, BALB/c SCID mice received colitis-inducing CD62L+CD4+ T cells and DX5+NKT cells as before, but also received anti-CD1d blocking antibody. Blocking of CD1d in the presence of DX5+NKT cells resulted in weight loss typical of chronic colitis in mice receiving only CD62L+CD4+ T cells (Fig. 3C). The clinical findings were confirmed histologically (Fig. 3D, E). These results suggest that activation of DX5+NKT cells by CD1d is necessary to prevent CD62L+CD4+ T cell-induced colitis.

DX5+NKT cells reduce established colitis

The clinical setting presents the problem that colitis is a condition requiring treatment after it has already been established. Therefore, we tested experimentally whether DX5+NKT cells can be used therapeutically to reverse established colitis in mice. In these experiments BALB/c SCID mice were induced with CD62L+CD4+ T cells to develop colitis, as usual. However, DX5+NKT cells (5 × 105) were not transferred to the mice until after 6 weeks, when colitis is normally apparent from weight loss measurements and histology (Fig. 2). DX5+NKT cell treatment was repeated on the 8th week after CD62L+CD4+ T cell transfer. After the first injection of DX5+NKT cells, mice began to gain weight, in contrast to the control and to the CD8+ cell-injected groups (Fig. 4A). Mice were killed after 10 weeks, and histology confirmed a small, but significant reduction in colitis in DX5+NKT cell-treated animals (Fig. 4B), versus controls. Furthermore, we investigated whether DX5+NKT cell therapy could treat established colitis in immunocompetent mice. For these experiments, DSS in drinking water was used to induce chronic colitis in BALB/c mice. Indeed, histological analyses showed a significant decrease in colonic inflammation in mice receiving DX5+NKT cells (Fig. 4C). Therefore, therapeutic application of DX5+NKT cells can lessen the degree of colitis in mice with established chronic colitis.

Figure 4.

DX5+NKT cells were tested for their ability to reverse established colitis. In these experiments, DX5+NKT cells (or CD8+ control cells) were injected 6 weeks after CD62L+CD4+ T cell transfer. (A) Weights of the mice in each group were recorded (n=6 mice/group, in each of two experiments). These results show that only mice treated with DX5+NKT cells were able to maintain their weight. A protective effect of DX5+NKT cells was also seen in the histological evaluation of colon specimens (B). (C) Finally, the effect of DX5+NKT cells was also tested in mice with established DSS-induced colitis in immune competent BALB/c mice (n=6 mice/group, in each of two experiments). DX5+NKT treatment reduced the histological signs of colitis. Moreover, simultaneous treatment with CD1d blocking antibody abrogated the DX5+NKT cell effect, confirming our observation in the CD62L+CD4+ T cell transfer model. p<0.05 when comparing groups indicated by*.

DX5+NKT cells reduce colitis cells through PD-L1 in vitro

In vitro assays were performed to investigate the interactions between DX5+NKT and CD62L+CD4+ T cells. The number of CFSE-labeled CD62L+CD4+ T cells decreased significantly after co-culturing with DX5+NKT cells for 3 days. Interestingly, CD62LlowCD4+ T cells co-cultured with DX5+NKT cells were only minimally affected with regard to cell number (Fig. 5A). Separation of the DX5+NKT cells from the CD62L+CD4+ T cells with a semi-permeable membrane inhibited the killing effect completely, suggesting the necessity for cell contact to kill the T cell targets (Fig. 5A). Because CD62L+CD4+ T and colitis cells express CD1d, the potential role of this molecule was investigated with anti-CD1.1 blocking antibody. Although not complete, blocking of CD1d did reduce the killing effect of DX5+NKT cells (Fig. 5B). Labeling CD62L+ cells with annexin V confirmed the ongoing process of cell death after co-culture with DX5+NKT cells (Fig. 5C). Recent publications indicate that PD-L1 is important for the induction of apoptosis of effector T cells during immune regulation 31. Therefore, we examined the possibility that PD-L1 expression is involved in the killing effect of DX5+NKT cells. Interestingly, not only do DX5+NKT cells express PD-L1 (Fig. 5D), blocking this molecule with specific antibody essentially abrogated the killing effect on target cells (Fig. 5B). One already known counterpart of PD-L1 is PD1, and notably, PD1 was expressed on both CD62L+ and CD62Llow cells (Fig. 5E), suggesting that PD-L1 killing activity is likely mediated via a different receptor. Therefore, these experiments suggest that DX5+NKT cells may serve to reduce colitis by deleting reactive lymphocytes through interactions involving CD1d and PD-L1.

Figure 5.

DX5+NKT cells have a killing effect on colitis-associated lymphocytes. (A) Three-day co-cultures of equal numbers (3 × 105) of DX5+NKT cells with CFSE-labeled CD62L+CD4+ T cells resulted in a significant reduction of CD62L+CD4+ T cell numbers (versus CD62L+CD4+ T cells cultured alone). This effect was completely abrogated when the two cell populations were separated in a transwell system, indicating the need for cell contact. Interestingly, CD62lowCD4+ T cell numbers were not affected by DX5+NKT cells. (B) Addition of CD1d blocking antibody slightly reduced the cytoreductive effects of DX5+NKT cells, and PD-L1 blocking nearly abrogated the effect. (C) CD62L+CD4+ T cells co-cultured with DX5+NKT cells showed annexin V positivity by flow cytometry analysis, indicating the target cells were experiencing cell death. (D) A role for PD-L1 in the killing effect of DX5+NKT cells is conceivable, since DX5+NKT cells express PD-L1 by flow cytometry. (E) One known receptor for PD-L1, referred to as PD-1, is expressed on both CD62L+CD4+ and CD62LlowCD4+ T cells. All values shown in the graphs are the mean ± SEM from three to eight independent experiments. p<0.05 when comparing two groups with matching symbols.


Progress is being made in the understanding of the pathogenesis of chronic colitis. Most studies suggest that an inappropriate immune response leads to destruction of epithelial and mucosal structures 32. One proposed mechanism of chronic colitis relates to a decrease or disturbance in immune regulation by T cells 33, 34. Consistent with this hypothesis, transfer of CD45RBhigh T cells in SCID mice induces colitis that can be prevented by co-transfer of the CD45RBlow T cell population, which contains regulatory T cells 1, 35. In this same respect, another study shows that colitis in Tgϵ26 mice, which develop colitis-inducing T cells, but not counter-balancing regulatory T cell populations, can be prevented by T cells that have undergone normal thymic development 36. Moreover, NKT cells can indeed regulate immune responses associated with autoimmune diseases. In the case of colitis, NKT cells activated through α-galactosylceramide can prevent acute inflammation of the bowel in mice 18. In the present study, we have demonstrated that DX5+NKT cells not only can prevent the development of colitis, they can lessen established chronic colitis progression in mice. Importantly, from a mechanistic perspective, our data suggest this suppressive effect against colitis is related to expression of CD1d on gut epithelium. Furthermore, our investigations reveal that the DX5+NKT cell-mediated effect is related to induction of colitis-associated lymphocyte cell death, which may be at least in part due to expression of the costimulatory-family molecule, PD-L1.

In our study we have hypothesized that DX5+NKT cells are stimulated in the gut by their counterpart molecule, CD1d. Interestingly, CD1d was highly expressed on CD62L+CD4+ T cells, and in addition, on epithelial cells of the colon after induction of colitis. Even 1 week after injection of CD62L+CD4+ T cells, CD1d was up-regulated in the colon without any sign of colitis. These findings correspond with the observation of Page et al. 37 who detected an increase in precipitable CD1d in inflamed colonic tissue of IBD patients. CD62LlowCD4+ T cells were also shown to express CD1d, but their transfer did not result in CD1d up-regulation on epithelial cells of the colon. More direct evidence of CD1d involvement came from experiments in which the inhibition of colitis by DX5+NKT cells, after colitis induction with CD62L+CD4+ T cells, was abrogated by CD1d blocking antibody. Our results suggest that the immune regulating function of NKT cells isolated from the spleen via DX5 and CD3 require activation through CD1d, bearing in mind that a recent publication indicates only a subset of DX5+NKT cells may be CD1d dependent 38. It is also notable that our results do not completely rule out the possibility that CD1d blocking antibody may also trigger CD1d, and thus in some way promote inflammatory responses in the gut, especially through CD1d ligation on monocytes, which could enhance inflammation via induction of IL-12 39. However, previously published data indicate that CD1d cross-linking on epithelial cells leads rather to the production of barrier-enhancing and immunoregulatory cytokines 40. Nevertheless, more in-depth experiments will be needed to exclude this alternative explanation for our results. At present, our results suggest a role for CD1d in the NKT cell-mediated suppression of gut inflammation.

Our in vitro co-culturing experiments showed that DX5+NKT cells induce CD62L+CD4+ T cell death, and this effect requires cell-cell contact and activation through CD1d. Interestingly, although CD62LlowCD4+ T cells also express CD1d, we did not observe a similar decrease in their numbers after co-culture with DX5+NKT cells. These data might suggest that the antigens presented by CD1d to DX5+NKT cells are different on CD62L+versus CD62LlowCD4+ T cells. Intracellular CD1d is transported through the endosomal reticulum and late endosomes, where lipid transfer proteins play a key role in loading the molecule 41. The detailed process of trafficking and loading has not yet been identified 42. Nevertheless, different lipids able to bind have been described, and recently, Zhou et al. 43 showed that NKT cell function requires the lysosomal glycosphingolipid isoglobotrihexosylceramide (iGb3), presented through CD1d. Interestingly, iGb3 seems to play an important role in the activation of NKT cells 44, especially during infections with bacteria that contain lipopolysaccharide (LPS) in their cell wall. Accordingly, NKT cells as regulator cells presumably require specific stimulation to affect immune-competent cells, and the lipid antigen presented by CD1d may play a pivotal role. Further experiments will be needed to better understand the selectivity of the DX5+NKT cell killing effect towards CD62L+CD4+ T cells.

Also with respect to the killing effect of DX5+NKT cells, we demonstrated the expression of PD-L1 on their cell surface. PD-L1 has recently been described as a member of the B7 family of costimulatory molecules, and appears to have a role in controlling the immune response, including the induction of apoptosis of tissue invading effector T cells associated with autoimmune disease 31. It has also been shown that PD-L1 may confer resistance to tumors, so that T cells directed against tumor cells are deleted when approaching the tumor 31. These studies, combined with our results showing strong PD-L1 expression on DX5+NKT cells, opened the question of whether this molecule was involved in the killing effect of DX5+NKT cells on lymphocytes. Indeed, blocking PD-L1 with specific antibody resulted in a reduced ability of DX5+NKT cells to induce the death of CD62L+CD4+ T cells and lymphocytes derived from intestinal lymph node tissue of mice with chronic DSS-mediated colitis. PD1, the one known receptor for PD-L1, was recently described as an inhibitor of T cell responses 31. However, our experiments show expression of PD1 on CD62L+CD4+ T cells, as well as on CD62LlowCD4+ T cells, suggesting the killing effect we observe may be via a receptor other than PD1. In fact, recent studies indicate that PD-L1 probably induces killing via a receptor other than PD1 45. Although this receptor has not yet been identified, it could be speculated that it is expressed on CD62L+CD4+, but not CD62LlowCD4+ T cells. These results, compared with published data that blocking of PD-L1 suppresses chronic colitis in the CD45RBhigh model in the absence of NKT cells, indicate that PD-L1 plays a key role in immune regulation depending on the cell population expressing this ligand 46.

In conclusion, our data show that DX5+NKT cells can play a role in down-regulating the effector immune response in mice that mediates colitis establishment and progression. Our results suggest that, for DX5+NKT cells to play a role in controlling intestinal inflammation, they require interaction with CD1d, presumably on intestinal epithelial cells. Once DX5+NKT cells are activated and expanded, our data suggest PD-L1 molecules expressed on their cell surface may be important for its cytoreductive action on activated lymphocytes associated with colitis. Future studies will be needed to better clarify the more precise mechanisms related to this hypothesis, and to determine whether DX5+NKT cells could be used as a form of cell therapy, or be targeted in vivo, as a means to treat chronic inflammatory bowel diseases.

Materials and methods

Cell harvesting and MACS

CD62L+CD4+ and DX5+NKT cells were purified from splenic mononuclear cells isolated from BALB/c mice (Charles River Laboratories, Wilmington, MA) by MACS. In a first step, splenic CD4+ cells were purified by negative selection, using a CD4+ T cell isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). Cells were passed through a MACS column (type LS) attached to a Midi-MACS-magnet (Miltenyi Biotec), according to the manufacturer's instructions. Unbound cells were collected in the negative fraction and labeled with a rat anti-mouse mAb against CD62L (clone MEL-14, BD Biosciences, Heidelberg, Germany). The cells were incubated with goat anti-rat IgG MicroBeads and magnetically separated. CD62L+(CD4+) cells were collected in the positive fraction, and CD62Llow(CD4+) cells in the negative fraction.

For DX5+NKT cell isolation, splenocytes were first depleted of non-T cells by labeling them with the following antibodies: rat anti-mouse CD11b-biotin (clone: M1/70), rat anti-mouse CD11c-FITC (clone: HL3), rat anti-mouse TER-119-biotin (clone: TER-119), rat anti-mouse CD45R-PE (clone: RA3–6B2) and rat anti-mouse Ly-6G-purified (clone: RB6–8C5) (all antibodies from BD Biosciences). Labeled cells were depleted using anti-rat-MicroBeads, anti-FITC-MicroBeads, anti-PE-MicroBeads and anti-biotin-MicroBeads, by passing them through a LS-separation column attached to a MidiMACS magnet (Miltenyi Biotec). T cells were collected in the negative fraction. To remove dead cells, the enriched T cell population was incubated with BasicMicroBeads and passed through a LS-separation column attached to a MidiMACS magnet. The negative fraction was labeled with anti-mouse-NK-MicroBeads (clone: DX5, Miltenyi Biotec) and passed through the usual LS-separation column attached to a MidiMACS magnet. The negative fraction was collected for purification of CD8+ cells, and DX5+NKT cells were collected in the positive fraction. The NK T cells were labeled with anti-mouse-CD8a-MicroBeads (Miltenyi Biotec) and a magnetic separation was performed as before, with CD8a+ T cells collected in the positive fraction. The purity of the resulting cell populations was tested by flow cytometry 47.

Colitis cells were isolated from lymph nodes of BALB/c mice after induction of colitis with DSS. Intestinal lymph nodes were disrupted and tissue was passed through a 40-µm cell strainer. Lymphocytes were collected for use in in vitro experiments.

The following antibodies were used for cell labeling: FITC-conjugated anti-mouse-CD4 (clone: GK1.5, rat IgG2b), PE-conjugated anti-mouse-CD62L (clone: MEL-14, rat IgG2a), FITC-conjugated anti-mouse-CD8a (clone: 53–6.7, rat IgG2a), FITC-conjugated anti-mouse CD3 molecular complex (clone: 17A2, rat IgG2b), PE-conjugated anti-mouse CD49b (clone: DX5, rat IgM) (all from BD Biosciences), PE-conjugated anti-mouse PD-L1 (clone: MIH5, rat IgG2a, eBioscience) and PE-conjugated anti-mouse PD1 (clone: RMP1–30, rat IgG2b, eBioscience).

Animals and cell transfer

To induce colitis, 1 × 106 CD62L+CD4+ T cells were injected i.p. into BALB/c SCID mice. DX5+NKT, or CD8+ control, cells (5 × 105) were subsequently i.p. injected into these mice on day -2, and at 2, 4 and 6 weeks after CD62L+CD4+ T cell transfer. In the experiment aimed at treatment of existing colitis, DX5+NKT or control cells were injected at 6 and 8 weeks after transfer (day 0) of CD62L+CD4+ colitis-inducing T cells. Six animals were included in each of the experimental groups. For CD1d blocking experiments, 50 µg (in 200 µL) CD1d-specific antibody (clone: Ly-38, Ig G2b, Pharmingen, BD) was injected i.v., which has been previously shown to block CD1d 48, 49. Every second day after cell transfer, mice were assessed for their overall condition and were weighed.

In some experimental groups chronic colitis was established with four cycles of 5% DSS (MW 40 000, ICN, Eschwege, Germany) dissolved in the drinking water. Each cycle consisted of a period of 7 days on DSS water, followed by normal drinking water for 10 days. These animals received DX5+NKT cells 4 weeks after the end of the last DSS cycle. Animal experiments were performed in accordance with the German legislation on the protection of animals.

Co-culturing experiments

CD62L+CD4+ T cells and DSS-colitis cells of BALB/c mice were labeled with CFSE (also called CFDA SE) with the Vybrant CFDA SE Cell Tracer Kit (Molecular Probes), according to manufacturers instructions. CFSE-labeled cells were co-cultured with an equal number of DX5+NKT cells (3 × 105) in 24-well plates for 3 days. In some cultures, the CFSE cells were physically separated from the DX5+NKT cell population with a trans-well insert. In other experiments, antibodies blocking CD1d (50 µg/mL, clone:1B1, IgG2b, Pharmingen, BD) and PD-L1 (50 µg/mL, clone: MIH5, IgG2a, eBioscience) were added to co-cultures. After the 3-day co-culture period, CFSE-labeled cells were counted with a fluorescence microscope and cells were further analyzed by flow cytometry. FACS staining for annexin V was performed with annexin V-FITC (BD Biosciences). Cells were labeled with anti-CD49b-PE (50 µL/mL, clone: DX5, rat IgM) for the identification of DX5+NKT cells in co-culture.


Standard hematoxylin and eosin staining was performed on colonic tissue to assess the degree of inflammation. The scoring was performed by a blinded observer, as previously described 50. Briefly, a score of 0–4 (4 being the most severe colitis) was assigned for epithelial loss and inflammatory infiltration. Mice were scored individually, with each value representing the mean score of three sections of the distal third of the colon.

Isolation of epithelial cells

Colonic epithelial cells were isolated as previously described 51, and tested for CD1d expression. Briefly, the colon was longitudinally incised and washed in PBS. The colon tissue was vigorously stirred at 37°C for 20 min in 20 mL Hanks’ balanced salt solution containing 1 mM EDTA. The resulting cell population was passed through a 40-µm cell-strainer, with epithelial cells being retained in the strainer. Epithelial cells were collected, and isolation of total RNA was performed using the RNeasy Kit (RNeasy Kit, QIAGEN, Hilden, Germany).


Real-time PCR for CD1d was performed on samples from healthy mice and mice with chronic colitis. Total RNA was isolated from epithelial cells, as described above, and 1 µg RNA from each sample was reverse transcribed in a total volume of 40 µL, using the RT system (Promega, Mannheim, Germany). Real-time PCR was performed using the LightCycler System and Fast-Start LightCycler–DNA Master SYBR Green I (Roche, Mannheim, Germany). Two microliters of the cDNA preparation were amplified with the following primer sets: CD1d (forward 5′-CTAGAGGCAGGGAAGTCAGA–3′; reverse 5′-GAGGCAGGTGTAAGGAAGAG –3′); and β-actin 52 (forward 5′-AGAGGGAAATCGTGCGTGAC–3′; reverse 5′-CAATAGTGATGACCTGGCCGT–3′). Conditions for the amplification were 95°C denaturation (5 s), with an annealing temperature of 62°C (5 s), and elongation at 72°C (20 s). The melting point for CD1d was 88.0–88.5°C, and for β-actin, 86–87°C. Aliquots of the samples were analyzed by gel electrophoresis and all PCR products were verified by DNA sequencing. For quantification, standard curves for CD1d and β-actin were made using agarose gel-purified, photometrically measured, PCR products. PCR was performed on tenfold dilutions of these purified products to construct standard curves. Comparisons for CD1d were only made between samples containing the same amount of β-actin.


CD1d expression on BALB/c SCID colon epithelial cells was examined by the immunoperoxidase method. Frozen colonic tissues were sectioned and labeled with a biotinylated mAb against murine CD1d (clone Ly-38). Sections were incubated with an avidin-biotin peroxidase complex (Vectastain Elite ABC reagent; Vector Labs). The peroxidase reaction was developed using 3,3′-diaminobenzidine tetrahydrochloride and all sections were counterstained with hematoxylin. A negative control reaction was performed using an isotype rat IgG2b mAb.


In vivo experiments in mice consisted of six animals per group, per experiment. Histology and cell count data are presented as the mean value ± SEM. Statistical analyses were performed using either a Student's t-test or the Mann-Whitney-U-test. Differences were considered significant at p<0.05.


The authors thank Karoline Edtinger and Manuela Kovacs for excellent technical assistance.


  1. 1


  2. 2


  3. 3


  4. 4


  5. 5