Inhibition of NKG2D receptor function by antibody therapy attenuates transfer-induced colitis in SCID mice



A role for the activating NK-receptor NKG2D has been indicated in several autoimmune diseases in humans and in animal models of type 1 diabetes and multiple sclerosis, and treatment with monoclonal antibodies to NKG2D attenuated disease severity in these models. In an adoptive transfer-induced model of colitis, we found a significantly higher frequency of CD4+NKG2D+ cells in blood, mesenteric lymph nodes, colon, and spleen from colitic mice compared to BALB/c donor-mice. We, therefore, wanted to study the effect of anti-NKG2D antibody (CX5) treatment initiated either before onset of colitis, when the colitis was mild, or when severe colitis was established. CX5 treatment decreased the detectable levels of cell-surface NKG2D and prophylactic administration of CX5 attenuated the development of colitis significantly, whereas a more moderate reduction in the severity of disease was observed after CX5 administration to mildly colitic animals. CX5 did not attenuate severe colitis. We conclude that the frequency of CD4+NKG2D+ cells increase during development of experimental colitis. NKG2D may play a role in the early stages of colitis in this model, since early administration of CX5 attenuated disease severity.


Crohn's disease


control Ig


inflammatory bowel disease


lamina propria lymphocytes


mesenteric lymph node


MHC class I chain-like gene A(B)


white blood cell


Inflammatory bowel disease (IBD) encompasses two related, chronic diseases of the intestinal tract, Crohn's disease (CD) and ulcerative colitis. It is thought that IBD is caused by an exaggerated immune response towards the normal, residing microbial gut flora 1, but neither causative agent(s), nor curative treatment strategies have unambiguously been demonstrated.

The NKG2D receptor is normally expressed on natural killer (NK) cells, CD8+ αβ T cells, and γδ+ T cells in humans 2, and on NK cells, activated macrophages, activated CD8+ αβ T cells, subsets of NKT cells, and γδ+ T cells in mice 35. Expression of NKG2D on CD4+ T cells has been reported in patients with CD, rheumatoid arthritis (RA), and HTLV-1-associated neurologic disease (M. Allez et al., personal communication and 6, 7), while this was not found in animal models of human disease 4, 8, 9. The receptor associates with the signaling adaptor peptide, DAP10 and engagement of the receptor-complex can trigger cellular cytotoxicity, cytokine production or proliferation, depending on the cell type 2, 10, 11.

In humans, ligands for NKG2D include the non-classical MHC class I chain-like gene A(B) surface molecules (MICA/MICB) and the UL16-binding proteins 10, 12, 13. The former are encoded by the MICA and MICB genes, which map to an area within the HLA class I region on chromosome 6 1416. Several studies have demonstrated susceptibility loci for IBD in the MHC region 17, 18 and a genetic association between IBD and MICA has been suggested 1821, although this is still controversial 22, 23. In non-pathological conditions, expression of MICA seems confined to the intestinal and thymic epithelia 15, 24, 25, but expression on other cell types can be induced by cellular stress, transformation, or infection 3, 5, 10, 15, 24, 26 and has been noted in autoimmune diseases such as celiac disease 27, rheumatoid arthritis 6, HTLV-1-associated neurologic disease 7, and CD (M. Allez et al., personal communication). In mice, the family of retinoic acid-early inducible gene-1 (RAE-1) molecules, the minor histocompatibility molecule H60, and murine UL16-binding protein-like transcript 1 (MULT-1) serve as ligands for NKG2D 3, 12, 28, 29. The normal cellular expression of RAE-1 and H60, generally corresponds to the expression of MICA in humans 28 while increased expression of RAE-1 was reported in the pancreas during development of autoimmune type 1 diabetes 8.

Despite the increased expression of NKG2D and its ligands in autoimmune diseases, the role of NKG2D-ligand interactions under these conditions has not been fully clarified. However, administration of an anti-NKG2D mAb (clone CX5) prevented the development of disease in a murine model of type 1 diabetes (T1D) 8 and prophylactic treatment in a murine model of multiple sclerosis with a second anti-NKG2D antibody (clone C7) reduced the severity of the disease 9.

As we detected expression of NKG2D on CD4+ T cells during development of colitis in a murine adoptive transfer model of IBD, we therefore wanted to assess whether treatment with anti-NKG2D (CX5) - prophylactic as well as after onset - could alleviate the disease. In this model, colitis development is uniform and occurs in all successfully reconstituted mice. Already by 2 weeks after transfer, multiple changes are observed in the colon wall, including (i) increased inflammatory cell content, (ii) increased cytokine levels in colonic homogenates (iii) hyperproliferation of the epithelium, (iv) loss of mucus content and (v) increased thickness with increased crypt length 30. These changes correlate with an increased ratio between the weight and the length of the colon, loose stools and presence of occult blood in the feces, while effects on body weight appears to be a slightly later manifestation of colitis. Thus, the rapid, reliable, and uniform colitis development makes this model highly suited for testing therapeutic as well as prophylactic protocols.

Although CX5 treatment could not prevent weight loss and development of colitis, mice treated with the CX5 mAb had significantly improved clinical parameters of the disease, including a decrease in weight-length ratio of the colon, colon wall thickness and histological score.

We therefore conclude that inhibition of NKG2D expression can ameliorate the severity of colitis in a murine adoptive transfer model.


Expression of NKG2D in healthy and colitic mice

The cellular expression of NKG2D on T cells and NK cells was tested in BALB/c mice and un-reconstituted C.B-17 SCID mice. In the peripheral blood and spleen of BALB/c mice, NKG2D was detected mainly on CD49b+ NK cells, whereas CD4+NKG2D+ cells in blood and spleen were extremely rare (Fig. 1A and data not shown). Similarly, in the mesenteric lymph nodes (mLN) of BALB/c mice, only very few cells were CD4+NKG2D+ (0.7 ± 0.2%, Fig. 1A). In un-reconstituted C.B-17 SCID mice, NKG2D was also only expressed on CD49b+ NK cells, but at slightly lower levels compared to BALB/c mice (data not shown). Thus, in healthy mice, we detected NKG2D primarily on CD49b+ cells and rarely on CD4+ T cells in blood, mLN, or spleen.

Figure 1.

(A-C) Representative FACS plots of NKG2D, CD4 and CD49b expression gated on live CD45.2+ cells. (A) BALB/c donor mice. NKG2D is expressed on CD4CD49b+ cells, but not on CD4+ cells. (B) Colitic SCID mice 3 weeks after reconstitution. NKG2D is expressed on a sub-population of CD4+ cells in addition to the CD4 cells. (C) Left: H60-transfected Ba/F3 cells stained with murine NKG2D-Fc or human GHR-Fc fusion. Low pH disrupted the binding of mNKG2D to H60. Right: Blood cells from reconstituted, colitic SCID mice stained as in (A). No significant difference in NKG2D staining before or after acid stripping. (D) Left: Percentage CD4+NKG2D+ cells in blood from BALB/c mice and reconstituted SCID mice 2–5 weeks after reconstitution. Weeks 2–3, untreated mice; weeks 4–5, animals treated with CX5 or cIg from week 2 to 4 or from week 3 to 5. Right graph: Percentage NKG2D+CD4+ cells in the mLN from BALB/c and colitic mice and the LPL from colitic mice. (E) NKG2D and HPRT RT-PCR on purified cell subsets of CD4+CD25 T cells and on CD4+ cells purified from mLN and spleen of colitic mice (20, 25, and 30 cycles of amplification). Spleen cells stimulated with Con A were used as positive controls.

Next, we studied the expression of NKG2D during the development of colitis in SCID mice reconstituted with CD4+CD25T cells from BALB/c donors. NKG2D expression on CD49b+ cells did not increase during the course of disease, rather we saw a small but significant decrease in mean fluorescence intensity of NKG2D (data not shown). Our most significant finding was an increased expression of NKG2D on CD4+ T cells in the peripheral blood, spleen, mLN and colon from mice developing colitis (Fig. 1B and D, and data not shown). CD4+NKG2D+ T cells appeared rapidly during the development of disease, and their frequency increased from 0.5 ± 0.1% (mean ± SEM) of live lymphocytes (equivalent to 12 ± 1.0% NKG2D+ of CD4+) in the peripheral blood at week 2 after adoptive transfer to 2.2 ± 0.3% at week 4 (equivalent to 7.8 ± 0.6% NKG2D+ of CD4+, Fig. 1D). A similar increase in CD4+ NKG2D+ cells could be detected in both mLN and among colonic lamina propria lymphocytes (LPL), where 5.7 ± 0.5% and 5.6 ± 0.4%, respectively, of the live CD45.2+ lymphocytes were CD4+NKG2D+ at week 3 (equivalent to 7.9 ± 0.6% and 7.2 ± 0.4% NKG2D+ of CD4+, Fig. 1B and D). Comparable frequencies were detected in mLN and LPL at week 4 (data not shown).

Soluble NKG2D may bind to cells expressing ligands for NKG2D, e.g. H60, and it was formally possible that the CD4+ cells detected during development of colitis were positive for NKG2D due to the ligand binding of soluble NKG2D rather than the cellular expression 31. To test this, we performed acid stripping 32 on blood samples from colitic SCID mice as well as on a H60-expressing cell line, which was incubated with NKG2D-Fc prior to acid treatment. These experiments showed that while the binding of NKG2D-Fc to H60 expressing cells was abolished after acid stripping, no loss of NKG2D expression could be detected on CD4+ cells from colitic BALB/c mice (Fig. 1C).

Furthermore, RT-PCR analysis on CD4+ cells purified from the spleen and mLN of colitic SCID clearly showed expression of NKG2D mRNA (Fig. 1E).

Together, these data strongly suggest that NKG2D is induced on CD4+ cells in mice during development of colitis.

Immunohistochemistry performed on frozen sections of the colon from un-reconstituted and reconstituted SCID mice with overt colitis showed a massive infiltration of CD4+ cells in the diseased animals and no CD4+ cells in the un-reconstituted mice (Fig. 2A and data not shown). In un-reconstituted mice, very few NKG2D+ cells were detected (Fig. 2B), whereas in reconstituted colitic mice NKG2D-expressing cells were detected in the lymphoid subepithelial structures as well as between the crypts in the lamina propria (Fig. 2C–E).

Figure 2.

Immunostaining for CD4 and NKG2D in frozen sections of colonic tissue from un-reconstituted controls and reconstituted colitic mice. (A) Massive infiltration with CD4+ cells (brown stain) in reconstituted colitic mice. (B) Insignificant NKG2D expression (red stain) in un-reconstituted control. (C) Accumulation of NKG2D+ cells (red stain, arrows) between the crypts in lamina propria of reconstituted mice. (D–E) High magnification of NKG2D+ cells. Isotype controls for both stainings were negative. Bars: A-C = 50 μm; D-E = 20 μm.

Effect of CX5 treatment on mice developing colitis – intervention study

We first studied whether the expression of NKG2D could be modulated by administration of CX5 mAb from weeks 2–4 or from weeks 3–5, respectively, after adoptive transfer of CD4+CD25 T cells, and how CX5 treatment affected the course of disease in the mice. Control mice were treated in parallel with an isotype-matched control Ig (cIg, rat IgG1). As demonstrated in Fig. 1D, essentially no CD4+NKG2D+ cells were detected after CX5 treatment whereas the cIg did not change the detected NKG2D levels.

When CX5 treatment was initiated after the mice had developed mild colitis (i.e. at week 2), no effect was seen on weight loss (Fig. 3A), or fecal consistency score (data not shown). However, the CX5-treated group had a significantly lower increase in the white blood cell (WBC) count than the cIg group, indicative of an attenuated inflammatory response (Fig. 3B). Furthermore, the post-mortem analysis showed that inhibition of NKG2D expression had a positive effect on the local colonic inflammation, since the colonic weight:length ratio was significantly lower in the CX5-treated group compared to the control group (Fig. 3C). In accord, the histological measurements of colonic wall thickness were significantly less in the CX5-treated group compared to cIg-treated animals (Fig. 3D). The total histological score (as described in Table 1) was, however, not significantly different between the two groups (Fig. 3E).

Figure 3.

Effect of CX5 on mild colitis. CX5 or isotype cIg (200 μg/mouse) was administered from 2 weeks after reconstitution until 3 days before sacrifice at weeks 4. (A) Percentage change in the body weight during the study. Open triangles = un-reconstituted controls, closed triangles = cIg, closed squares = CX5. Error bars = SEM. (B) WBC count in controls (comparable at weeks 2 and 4) and reconstituted mice treated with either CX5 or cIg. Samples from week 2 were obtained prior to the beginning of treatment. Bars: Mean. * p <0.02. (C) Colonic weight:length ratio (mg/cm). Bars: Mean. * p <0.05. (D) Colonic wall thickness determined on histology slides by image analysis. Bars: Mean. * p <0.05. (E) Histological grading of the colonic tissue according to Table 1. Bars: Median.

Table 1. Histopathological grading system (maximum score is 13). Modified from [30]30, 38
Severity of inflammation0None
 1Mild mononuclear infiltration
 2Marked infiltration with lymphocytes and granulocytes Focal degeneration of crypts
 3Severe infiltration of inflammatory cells, multifocal crypt degeneration and/or erosions
Extent of inflammation0None
Amount of mucin0Normal
 1Slightly decreased
 2Moderate decrease Focal absence of mucus
 3Severe depletion of mucus Very few spots with mucus left
 4No mucus left
Degree of proliferation0None
 1Mild increase in crypt height
 2Moderate increase in crypt heightFocal marked increase
 3Marked increase in crypt height — entire section

When CX5 treatment was initiated in mice with fully developed colitis (at week 3) no effect on the course of disease during the 2-week observation period could be detected (data not shown). Thus, when intervening early in the course of disease (i.e. when the inflammatory changes are mild), administration of CX5 seemed to attenuate the escalation of the inflammation, but when the inflammation and tissue destruction was overt, there was no effect of the antibody.

Effects of CX5 treatment on mice developing colitis – prevention study

Since the outcome of CX5 treatment seemed to depend on the severity of disease, we tested the effect of treatment initiated before the onset of disease (i.e. preventive treatment). CX5 was administered to the mice from the day after transfer until 3 days before sacrifice at week 3. CX5 treatment did not prevent the development of colitis as loose stools and occult blood in feces were detected in both groups 2 weeks after transfer. However, whereas the fecal consistency score in the vehicle-treated group increased at week 3, this score decreased in the CX5-treated group and was significantly less than in the vehicle group at sacrifice (Fig. 4B). The scores for occult blood in feces at week 3 were comparable between the two groups [PBS vs. CX5: 2.5 (0–4) vs. 3 (0–4)], as were the body weight changes (Fig. 4A). WBC counts were significantly lower in the CX5-treated group (Fig. 4C) and post-mortem examination showed a significantly lower colonic weight:length ratio of CX5-treated mice, confirming the clinical findings (Fig. 4D). This was also reflected in the histological analysis of the colon, which showed a significantly thinner gut wall and lower histological score in the CX5-treated group (Fig. 4E–G).

Figure 4.

Effect of CX5 administered at the time of reconstitution. (A) Percentage change in the body weight during the study. Open circles = un-reconstituted controls, closed triangles = cIg, closed squares = CX5. Error bars = SEM. (B) Fecal score at weeks 2 and 3. Bars: Median. * p <0.02. (C) WBC count at the time of sacrifice. Bars: Mean. ** p <0.01. (D) Colonic weight:length ratio (mg/cm). Bars: Mean. ** p <0.006. (E) Colonic wall thickness (mm). Bars: Mean. * p <0.02. (F) Histological grading of the colonic tissue according to Table 1. Bars: Median. * p <0.05. (G) Representative photomicrographs of H&E/Periodic Acid Schiff-stained colonic tissue from un-reconstituted (left), reconstituted CX5-treated (middle), and reconstituted PBS-treated (right) mice. Bars = 100 μm.


In the present report, we have assessed the effect of CX5 administration in different stages of disease in a murine adoptive transfer model of colitis.

When CX5 was administered to animals with mild colitis, there was a moderate attenuation of the disease, whereas the antibody could not alleviate signs of colitis when administered to severely colitic animals, (despite that we were unable to detect any NKG2D expression on CD4+ T cells in CX5-treated mice by flow cytometry). Considering the fairly low frequency of NKG2D-expressing inflammatory cells, and the vast redundancy in the inflammatory cascades and signaling pathways that are activated during the development of colitis, it is not surprising that functional inhibition of NKG2D is insufficient to completely inhibit the inflammatory process, when this is already established and severe. Nevertheless, CX5 treatment attenuated colitis in the prevention study as well as in the early intervention study, and considering the robustness of this model and the measured biomarkers, we believe that our observations support the conclusion that the NKG2D receptor plays a role during development of colitis. Treatment with CX5 antibody reduced both the severity of colonic inflammation and the degree of epithelial hyperproliferation when administered in mild stages of the disease. Increased proliferation occurs early in the course of colitis and - together with the consequential loss of protective mucus - markedly reduces the intestinal barrier towards the luminal microbial flora 30. Thus, it seems imperative to avoid or reduce the proliferation before loss of the barrier. In this respect, it is noteworthy that CX5 treatment from the time of transfer led to a markedly reduced epithelial proliferation, possibly due to suppression of cytokine secretion, which is known to stimulate epithelial proliferation 33, 34. We also found a pronounced effect of CX5 treatment on the WBC count. Specifically, early intervention (Fig. 3B) prevented the rise in WBC count observed in cIg-treated mice. As CX5 is known to be a non-depleting antibody that mediates internalization of the NKG2D receptor 8 (and our own unpublished data), we suggest that cells relying on the NKG2D receptor for their activation and pro-inflammatory function are impeded by this treatment.

Other studies have shown that inhibition of NKG2D function may ameliorate autoimmune diseases. Thus, some inhibition of demyelination of nervous tissue in a murine model of multiple sclerosis was observed after administration of anti-NKG2D antibodies. In this model, only a fraction of the γδ-cells in the CNS expresses NKG2D 9. In the NOD mouse model of type 1 diabetes (T1D), complete protection against hyperglycemia was obtained after CX5 administration when treatment was initiated before the onset of diabetes, but after the initiation of insulitis 8. The complete protection observed in the latter may be due to a markedly higher expression of NKG2D on infiltrating CD8+ T cells, and thus a more pronounced pathogenic role of NKG2D in this model, but may also reflect that severe infiltration of the islets of Langerhans and destruction of up to 90% of the functional β-cell mass may be tolerated without overt hyperglycemia 35.

How inhibition of NKG2D function leads to attenuation of colitis, and whether inhibition on the different NKG2D-expressing cell types is of equal importance is not clear. It has previously been shown for human CD4+CD28NKG2D+ cells that stimulation through NKG2D with mAb or MIC enhanced anti-CD3-triggered cell proliferation and cytokine release 6, and comparable findings were done with CD8+CD28NKG2D+ cells 15. Recently, it was shown that intestinal epithelial cells from CD patients (which overexpress MICA) induced expansion of CD4+NKG2D+ cells, and that this expansion was inhibited when NKG2D-MICA interaction was prevented by a MICA mAb (M. Allez et al., personal communication). Interfering with NKG2D function in vivo using the CX5 antibody also inhibits proliferation of autoreactive T cells in the NOD mouse (Supporting Information in 8), and we find it likely that, also in our model, CX5 inhibits the costimulatory effects of the NKG2D receptor, thereby decreasing the proinflammatory potential of the CD4+NKG2D+ T cells. In non-pathological conditions, NKG2D is expressed on murine NK cells and γδ+ T cells, whereas the expression of NKG2D on CD4+ T cells is correlated to disease. We therefore speculate that in this colitis model, functional inhibition of NKG2D on the CD4+ cells plays the most significant role in attenuating the inflammatory response. While expression of NKG2D on CD4+ T cells was not detected in the previously studied animal models 4, 8, 9 it has, nevertheless, been detected on CD4+ T cells from patients with CD, RA, and HTLV-1-associated neurologic disease (M. Allez et al., personal communication, and 6, 7 ). The presence of CD4+NKG2D+ in this colitis model thus reflects the findings in CD, and may be augmented by the highly imbalanced composition of lymphocytes in the reconstituted SCID mice, since the adoptively transferred CD4+ T cells undergo a massive expansion, which may drive expression of NKG2D 36.

NKG2D expression on CD4+ T cells was readily detected during development of colitis and the levels of expression were comparable to those found on CD49b+ cells in normal BALB/c mice (Fig. 1) and on CD8+ T cells in the NOD mouse 8. However, following administration of CX5, NKG2D expression on CD4+ T cells was no longer detectable. Previous studies have shown that CX5 is a non-depleting antibody, which blocks binding of NKG2D to its ligands and mediates internalization of the receptor 8. We have confirmed these data but could also show that although most of NKG2D expression was lost after CX5 treatment, low levels of CX5-antibody could be detected on the cell surface using secondary antibodies specific for the Fc portion of CX5 (see Supporting Information Fig. 1). Still, regardless of whether CX5 mediates internalization or mask the NKG2D receptor on the cell surface (or both) NKG2D function is in any case blocked by CX5 treatment.

In conclusion, our study demonstrates an increased frequency of CD4+NKG2D+ T cells in SCID mice developing colitis. Administration of the NKG2D antibody, CX5 inhibited NKG2D expression and could ameliorate – but not completely prevent – development of colitis. These findings suggest a role for NKG2D in the early stages of disease and future studies should address whether this can be exploited therapeutically.

Materials and methods


Eight- to ten-week-old C.B-Igh-1b/IcrTac-Prkdcscid (C.B-17 SCID) and BALB/cAnNTac female mice bred under SPF conditions were purchased from M&B Taconic (Denmark) and housed under conventional conditions at Novo Nordisk a/s. No pathogens were detected in the housing facility during the study (testing as recommended by The animal studies were approved by the Danish Animal Experimentation Inspectorate.

Purification and adoptive transfer of CD4+CD25 T cells

Splenocytes from MHC-compatible BALB/c donor mice were positively selected for CD4+ T cells and depleted of CD4+CD25+ cells as previously described 30 (> 93% of the cells were viable and CD4+, and >98% were CD25). The C.B-17 SCID recipients were reconstituted with 300 000 cells by i.p. injection. FACS analysis of a peripheral blood sample 2 or 3 weeks after reconstitution confirmed the presence of CD4+ T cells in all recipients, indicating a successful reconstitution.

Isolation of LPL

LPL were isolated as described in [37]. After sacrifice, the colon was removed from diseased animals 3 or 4 weeks after reconstitution (n = 8 per week, pooled in tubes containing the colon from two-to-three mice), cut into small pieces and incubated in PBS with 2 mM EDTA for 20 min at 37°C. The samples were then washed in PBS and incubated for 90 min at 37°C in RPMI1640 medium containing 1 mg/mL Clostridiopeptidase A (Sigma) and 25 μg/mL DNase I (Roche). After washing and preparation of a single-cell suspension, the lymphocytes were isolated from the samples using a Lympholyte®-M density separation medium (Cederlane Laboratories, Canada) according to the manufacturer's instructions and the cells were then stained as described below.

Flow cytometric analysis of NKG2D expression

The expression of NKG2D in the peripheral blood, spleen, mLN, and LPL was evaluated by flow cytometry of samples from BALB/c donor mice, un-reconstituted (n = 5, blood and spleen only), and reconstituted SCID mice. The mice were under general anesthesia during blood sampling and were sacrificed prior to collection of spleen, LPL and mLN. Blood samples from the reconstituted mice were analyzed from 2 to 5 weeks after reconstitution, i.e. from when the first clinical signs of colitis occurred, and the CD4+ donor cells became detectable in the blood, to well after the colitis was fully developed 30. Spleen cells, mLN cells and LPL from diseased animals were analyzed 3 and 4 weeks after reconstitution. The samples were stained with 0.25 µg PE-conjugated anti-NKG2D (clone CX5, Nordic BioSite AB, Sweden) per 200 000 WBC or PE-conjugated anti-rat IgG1-isotype control (clone R3–34, BD PharMingen, Belgium) and co-stained with FITC-conjugated anti-CD45.2 (clone 104, eBioscience, CA, USA), biotin-conjugated anti-CD49b (clone DX5, BD PharMingen), and PerCP-conjugated anti-CD4 (clone L3T4, BD PharMingen) followed by streptavidin-APC. After lysis of red blood cells in BD PharmLyse™ lysis buffer, the samples were analyzed on a FACS Calibur or a BD™ LSRII flow cytometer using either CellQuest or DiVa software (BD Biosciences, CA, USA) for analysis. At least 5000 events were acquired within the live cell gate (based on forward- side scatter and expression of CD45.2).

Acid stripping

Acid stripping was performed to remove any NKG2D bound to its ligands on the cell surface 32. The mNKG2D-Fc fusion protein (2 μg, R&D Systems) was added to a H60 expressing Ba/F3 cell line 28 (kindly provided by Dr. L. Lanier, UCSF) and incubated for 30 min followed by washing in PBS. The cells were then incubated for 2 min in RPMI 1640, pH 4 or pH 7 prior to staining with PE-conjugated goat anti-human secondary F(ab′)2 antibody (Jackson ImmunoResearch Laboratories) and FACS analysis. The hGRH-Fc protein (2 μg, R&D Systems) incubated with H60 transfectants in a similar fashion served as a negative control. Three independent experiments were performed. In parallel, blood samples from colitic mice (n = 14) were incubated in RPMI buffer, pH 4 or pH 7 prior to staining with the following antibodies: FITC-CD45.2, PerCP-CD4 and PE-CX5. This was followed by FACS analysis, where the percentage of CD45.2+CD4+NKG2D+ cells among all CD45.2+CD4+ cells was calculated and a pair-wise comparison of the samples (incubated at pH 7 and pH 4, respectively) was done.

RT-PCR expression analysis

The expression of NKG2D transcripts was analyzed in CD4+ cells and CD4+CD25 cells of BALB/c donor mice (n = 14 and 6, respectively for the two cell subsets) and colitic reconstituted SCID mice (n = 30). Briefly, mLN and spleens were removed, and CD4+ T cells isolated using CD4-microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) followed by sorting on an AutoMACS (Miltenyi Biotec). For the colitic mice, the samples were pooled in two fractions containing cells from 15 animals each, while for CD4+ cells from BALB/c mice, two cell pools each containing cells from 6 and 8 mice, respectively, were prepared. FACS analysis showed that the cells were >97% CD4+ and that no CD4CD49b+ (representing NK cells) were present (data not shown). DNA from the isolated cells was analyzed by PCR using the following primers; 5′-TGTGGCTTGCCATTTTCAAAGAGACG-3′ and 5′-TTACACCGCCCTTTTCATGCAGATG-3′ (expected band size 446 bp). The NKG2D band was excised, purified and sequenced in order to confirm that the primers amplified NKG2D. As an internal control we used the primers for the Hprt-gene (Hypoxanthine guanine phosphoribosyl transferase), 5′-GTTGGATACAGGCCAGACTT TGTTG-3′ and 5′-AGAGGGTAGGCTGGCCTATAGGCT-3′ (expected band size 351 bp).


Samples of transverse colon from un-reconstituted and reconstituted colitic SCID mice were embedded in TissueTek® O.C.T. compound (Sakura, PA, USA), snap-frozen and stored at –80ºC. NKG2D and CD4 expression was subsequently assessed on frozen sections (8 µm thick). The sections were defrosted in 4% paraformaldehyde at 4°C, and endogenous biotin blocked by incubation with avidin- and biotin-blocking solutions (Dako a/s, Denmark). Nonspecific binding was blocked by incubation with TBS containing 3% skimmed milk (DifcoTM, BD Bioscience) and 10% goat serum (Dako a/s), followed by incubation with either rat anti-mouse CD4 (clone L3T4, BD PharMingen) or rat anti-mouse NKG2D (clone 191004, R&D Systems). Sections were then incubated with biotinylated goat anti-rat IgG (Jackson ImmunoResearch Laboratories, PA, USA) followed by a peroxidase-conjugated Avidin-Biotin complex (HRP-ABC) (Vectastain ABC peroxidase kit, Vector Laboratories, CA, USA). Sections were then blocked in Du Pont Blocking Reagent (TNB) followed by incubation with biotinylated Tyramide (TSA indirect, NEL 700) according to the manufacturer's recommendation (NEN, Perkin Elmer Life Science, MA, USA). Then, sections were incubated with either HRP-ABC for CD4 demonstration or alkaline phosphatase-conjugated Avidin-Biotin complex (Vectastain ABC Alkaline Phosphatase kit, Vector Laboratories) for demonstration of NKG2D. The chromogenic reaction was achieved with either Diaminobenzidin (Sigma) for CD4 assessment or Liquid Permanent Red (Dako a/s) for NKG2D assessment and counterstained with hematoxylin. CD4 stained sections were dehydrated and mounted in Pertex (Histolab AB, Sweden) whereas NKG2D-stained sections were mounted in aqueous Faramount Aqueous Mounting Medium (Dako a/s). Negative controls were omission of primary antibodies and incubation with either rat monoclonal IgG2a or IgG2b isotype-specific control antibodies (clones R35–95 and A95–1, respectively, BD PharMingen).

Effect of anti-NKG2D on colitis

Intervention study

We assessed the effect of anti-NKG2D (clone CX5, eBioscience) administration at different stages of disease by initiating treatment either 2 weeks (mild colitis) or 3 weeks (overt colitis) after adoptive transfer. The doses and dosing intervals were similar to those, which prevented diabetes in the study of Ogasawara et al. 8. In groups 1–4 (each group with n = 10), animals were dosed i.p. twice weekly with 200 μg antibody for 2 weeks according to the following scheme: Group 1: CX5 from week 2 to week 4 after transfer. Group 2: control rat IgG1 isotype (cIg) from week 2 to week 4 after transfer. Group 3: CX5 from week 3 to week 5 after transfer. Group 4: cIg from week 3 to week 5 after transfer. Group 5 (n = 5): un-reconstituted SCID mice receiving no treatment. The animals were sacrificed 3 days after last treatment. No mice died during the study except one mouse in group 2, which was sacrificed due to peritonitis following the first injection, and was excluded from the study.

Prevention study

In this study, we assessed if anti-NKG2D administered from the time of transfer could influence the occurrence or onset of disease. The animals were treated i.p. with either 200 μg CX5 or PBS twice weekly from the day after reconstitution to 3 days before sacrifice at the end of week 3 (n = 10). Five un-reconstituted mice received no treatment. No mice died before the end of the study period.

Assessment of colitis in vivo

Mice were weighed twice weekly, and sacrificed if they lost more than 20% of their initial body weight. Fecal consistency was evaluated before the start of treatment and at sacrifice using a semi-quantitative score (normal stool = 0; slightly soft = 1; soft but formed = 2; not formed = 3; liquid stools or no feces in colon at sacrifice = 4) as previously described 30. Presence of occult blood in feces was determined using a semi-quantitative scoring system based on the color intensity developing on the test plate after application of a fecal sample and test reagent (FecaTwin® test kit (Vitaltech Ibérica S.L., Spain) as previously described 30.

White blood cell count

Samples (20 µL) of EDTA-stabilized peripheral whole blood were analyzed in a Medonic CA 620 (Boule Nordic, Denmark) blood analysis apparatus according to the manufacturer's instructions, and the values expressed as the number of WBC/L.

Post-mortem procedure

Prior to sacrifice, the mice were anesthetized and blood from the peri-orbital venous plexus was collected in EDTA-containing Microvette® containers (Hounisen, Denmark) for FACS analysis and WBC count. The samples were kept on ice until analysis. After sacrifice, the colon was excised, rinsed gently with saline, and the weight and length recorded. The transverse colon was opened longitudinally, mounted on a plastic plate and fixed overnight in 4% paraformaldehyde, processed and embedded in paraffin. Where indicated, the spleens were removed for FACS analysis.


One paraffin-embedded section (∼7 µm) of the transverse colon from each animal was stained with H&E/Periodic Acid Schiff and analyzed by light microscopy. A total histological score was calculated for each animal according to Table 1. Measurements of the gut wall thickness (from the tunica serosa to the lamina epithelialis) were performed on all sections using Image Pro 4.1 (Media Cybernetics, MD, USA). The histological analyses were performed without knowledge of previous treatment of the mouse.

Statistical analysis

Fecal consistency score, occult blood score, and histological score are shown as median (range) and analyzed using the Mann-Whitney U-test. FACS data, WBC count, body weight, colonic weight:length ratio and gut wall thickness are shown as mean ± SEM and analyzed by Student's t-test using Welch's correction for unequal variances. Differences were considered statistically significant when p <0.05.


Stine Kjellev was supported by a co-financed PhD fellowship from the Danish Ministry of Science, Technology, and Innovation and the Corporate Research Affairs at Novo Nordisk a/s. We wish to thank Animal Unit at Novo Nordisk a/s, Gentofte and Måløv for excellent technical assistance. The technical assistance of Trine Larsen, Camilla Frost Sørensen, Heidi Winther, Steen Kryger, Rose Andresen, Stine Bisgaard, Daniel Højgaard and the photographical assistance of Grazyna Hahn is warmly acknowledged. We thank Dr. Lewis Lanier for providing the H60 expressing Ba/F3 cell line and for helpful discussions, and Dr. M. Gad for help and advice.

Note added in proof

The manuscript by Allez et al. referred to in the text as “personal communication” has now been accepted for publication. The reference is


  1. 1


  2. 2


  3. 3


  4. 4