Colitis-associated colorectal cancer driven by T-bet deficiency in dendritic cells

Abstract

We previously described a mouse model of ulcerative colitis linked to T-bet deficiency in the innate immune system. Here, we report that the majority of T-bet(-/-)RAG2(-/-) ulcerative colitis (TRUC) mice spontaneously progress to colonic dysplasia and rectal adenocarcinoma solely as a consequence of MyD88-independent intestinal inflammation. Dendritic cells (DCs) are necessary cellular effectors for a proinflammatory program that is carcinogenic. Whereas these malignancies arise in the setting of a complex inflammatory environment, restoration of T-bet selectively in DCs was sufficient to reduce colonic inflammation and prevent the development of neoplasia. TRUC colitis-associated colorectal cancer resembles the human disease and provides ample opportunity to probe how inflammation drives colorectal cancer development and to test preventative and therapeutic strategies preclinically.

Figure 1. TRUC mice develop colitis-associated colorectal cancer

A. % of TRUC mice with and without dysplasia and cancer is shown as a function of mouse age (mos). The number of mice surveyed is displayed along the x-axis. No cancers were observed in age-matched RAG2^−/− (data not shown). B. Dysplastic grade is shown as a percentage of dysplastic cases for mice at 3, 4, and 6 months. Size (mm) of low and high grade dysplastic lesions is displayed for each animal. C. Level of invasion (intra-mucosal vs sub-mucosal) is shown as a % of carcinoma cases. Lesion size (mm) of the carcinomas (shaded bars) and continuous areas of dysplasia (open bars) are plotted for individual cases. D. Photographs of TRUC colon with colitis but no dysplasia (left) and TRUC colon with colorectal cancer (right). E. Histology from TRUC mice with invasive carcinoma (upper panel) dysplasia (middle panel, (left: high grade dysplasia, right: low grade dysplasia), no dysplasia (lower panel, (left: colitis, no dysplasia, right: no colitis, no dysplasia), scale bars 100 microns. F. Immunohistochemistry. PCNA: (left upper panel), scale bars 100 microns. β-catenin (right upper panel) staining in dysplastic epithelial crypts (left) and nuclear staining in invasive carcinoma (right), scale bars 25 microns. COX-2: (left middle panel) and quantitation of COX-2+ crypts (right middle panel), scale bards 100 microns. P53 (left lower panel, nuclear staining in crypts) and quantitation of P53 staining (right lower panel), each dot represents data from one mouse, scale bars 25 microns. G. Colonic epithelial p53 activity induced by doxorubicin. (Left upper panel) Consensus oligonucleotide binding activity of p53 in nuclear extracts of colonic epithelial cells from RAG2^−/− vs TRUC non-neoplastic vs TRUC neoplastic colons. Gray bars are from untreated samples, black bars are from doxorubicin treated CECs incubated with biotinylated oligonucleotides, and open bars are from doxorubicin treated CECs incubated with biotinylated and 20-fold excess unbiotinylated oligonucleotides. Data represent the mean of two independent experiments with primary epithelial cells pooled from 20 mice per group (RAG2^−/−, TRUC non-neoplastic, and TRUC neoplastic), error bars denote +/− SD. (Left lower panel) p21 induction in response to doxorubicin. Real time qPCR was performed on control and doxorubicin treated samples. Fold change between doxorubicin and untreated samples was calculated by dividing their respective 2^−ΔCt values. (Right lower panel) APAF1 induction in response to doxorubicin was determined on the same sample set as above. Bars represent the mean of two independent experiments and error bars the +/− SD.

Figure 2. Chromosomal instability and reactive oxygen induced DNA adducts in TRUC carcinogenesis

A. Flow cytometric aneuploidy analysis of epithelial cells isolated from mice w/o dysplasia, w/dysplasia, and w/carcinoma. (Left panel) Representative plot from one mouse with diploid and aneuploid populations labeled. (Right panel) % of aneuploid epithelial cells for all mice analyzed. Each dot represents data from one mouse. Horizontal bars represent the mean. B. Reactive oxygen species were measured in TRUC vs RAG2^−/− distal colons. Reactive luminescence measured in units/sec/mg colonic tissue X 1000 (y-axis) is plotted as a function of mouse age. Open bars (RAG2^−/−), shaded bars (TRUC), 5–8 mice per group, means are graphed, error bars represent +/− SD, p<.001 for all comparison between TRUC 1 mos and TRUC 2–6 mos. C. 8’hydroxy-2’deoxyguanine (ng/ml) levels in colonic epithelial cells. RAG2^−/− control mice (age 1 and 6 mos, open bars, n.d. = not performed for RAG2^−/− age 2–4 mos) and TRUC mice (age 1–6 months, shaded bars). 6–9 mice were used per time point. Means are graphed; error bars represent +/− SD, p<.001 for all comparison between TRUC 1 mos and TRUC 2–6 mos. D. ELISA based protein determinations for selected cytokines are shown from explant cultures pooled from 5–7 specimens per time point from RAG2^−/− vs TRUC colitic (non-neoplastic) vs TRUC dysplastic vs TRUC carcinoma. Bars represent the mean values for three independent sets of pooled samples and error bars represent the +/− SD.

Figure 3. Imbalance between cell death and proliferative repair in TRUC mice

A. The TRUC epithelium is highly proliferative and epithelial proliferation increases with time. Cell proliferation index (BrDU+ cells/(epithelial cells/crypt)) normalized to RAG2^−/− levels is plotted along the y-axis. The mean of 6 RAG2^−/− mice (age 1 month) and 5–7 TRUC mice per time point (age 1–6 months) is shown, error bars represent +/− SD. B. Representative immunofluorescence micrograph of data quantitated in Panel A showing a hyperproliferative TRUC mucosa with BrDU staining (red) and DAPI-labeled nuclei (blue), scale bar 25 microns. C. Cell proliferation outpaces cell death as TRUC mice age. Sections from the mice in (A) were stained with the TUNEL reagent to label dying epithelial cells. The number of TUNEL+ epithelial cells/(epithelial cells/crypt) was then divided by the Cell Proliferation Index. This ratio normalized to the calculated value observed for the RAG2^−/− samples is plotted along the y-axis. The mean value for 6 RAG2^−/− mice (age 1 month) and 5–7 TRUC mice per time point (age 1–6 months) is shown, error bars represent +/− SD. D. Epithelial cell lysates were generated from distal colonic epithelial cells from TRUC and RAG2^−/−. Each lane represents sample from two mice and two separate groups with the indicated pathology were examined by Western blot for Bcl-2, Bak, Bax, Bcl-XL levels. Hsp 90 was used as a loading control.

Figure 4. TRUC colonic neoplasia is dependent upon an inflammatory response to commensal bacteria that is MyD88-independent and can be blunted by early but not late TNF-α neutralization

A. Six month TRUC mice treated with continuous broad spectrum antibiotics do not develop neoplasia. Histologic colitis score (y-axis) and neoplastic classification (x-axis) [low grade dysplasia (LGD), high grade dysplasia (HGD), and adenocarcinoma (ACA)]. Vancomycin, metronidazole, neomycin, and ampicillin treated (open circles) and water control (shaded circles). B. MyD88^−/− TRUC develop colitis and neoplasia. (Left panel) Histologic colitis scores for 8 week old MyD88^−/− TRUC (open circles) and TRUC (shaded circles). (Right panel) Histologic colitis scores (y-axis) and distribution of the degree of neoplasia (x-axis) for 12 week old MyD88^−/− TRUC. C. Mice treated with anti-TNF-α from 8–12 weeks do not develop neoplasia while isotype control 8–12 week group mice do. % of the population with no dysplasia, LGD, HGD, and ACA is shown for each group (n= 8 per group). D. Treatment with anti-TNF-α does not ameliorate colitis or prevent the development of neoplasia in mice treated from 3–6 months, histologic colitis score (y-axis) and distribution of the degree of neoplasia (x-axis) (open circles, anti-TNF-α treated; shaded circles, isotype control treated). Each circle represents an individual mouse).

Figure 5. TRUC colitis is dependent upon DCs and T-bet over-expression in DCs reduces the prevalence of TRUC colitis-associated dysplasia and colorectal cancer

A. TRUC colitis is dependent upon DCs. CD11c-T-bet-DTR-GFP tg TRUC mice were generated and used as donors in bone marrow chimera experiments with TRUC mice as recipients, TRUC bone marrow was engrafted into irradiated TRUC recipients as a control. After engraftment (8 wks), CD11c Tbet tg TRUC mice were treated every other day with PBS (shaded circles) or DTX (open circles) for 8 wks (left panel). Control TRUC bone marrow chimeras were DTX treated for 8 weeks (shaded squares) (left panel). Each dot represents one mouse. 40% of the original DTX-treated cohort group died during the 8 wk course of treatment. Deletion of CD11c+ DC subsets was confirmed by indirect immunofluorescence microscopy for CD11c on paraffin embedded sections from representative mice from each group. A selected image from the distal colon is shown for each group CD11c (pink), Hoescht (blue), scale bars 50 microns. B. T-bet expression in CD11c T-bet Tg TRUC mice. T-bet expression was measured in purified myeloid colonic cell populations, in colonic epithelial cell populations, and in splenic cell populations pooled from 15 sex-matched CD11c T-bet Tg TRUC mice without evidence of colitis, using real time qPCR. Bars represent the mean values across three replicate qPCR reactions on the samples and error bars +/− SD. C. Restoration of T-bet expression using a CD11c transgenic cassette reduces colonic explant TNF-α levels in transgene expressing TRUC mice. TNF-α ELISA-based protein determinations from colonic explants from CD11c T-bet Tg TRUC and TRUC, age 4 weeks. Bars represent the mean value across 3 sets of explant samples (5 individual mice per group) and error bars +/− SD. D. T-bet over-expression inversely correlates with colitis score. qPCR for T-bet was performed on cDNA generated from RNA isolated from paraffin embedded CD11c T-bet IRES GFP tg TRUC colons and RAG2^−/− controls. T-bet abundance relative to β-actin and then normalized to RAG2^−/− levels (y-axis) versus colitis score (x-axis). (Left panel) individual mice depicted by dots (when overlapping, the number of mice is shown above the dot. (Right panel) Bars represent the mean values for mice grouped by colitis score: (0), (1–3), (4–6), (7–9) and error bars represent +/− SD. E. CD11c T-bet tg TRUC mice have reduced prevalence of dysplasia and carcinoma. Cohorts were aged to six months and evaluated for colitis (y-axis) and neoplasia (grouped by histopathology along the x-axis), CD11c T-bet Tg TRUC (shaded circles, n= 35) and control TRUC (open circles, n= 48).