Wild-type and Cdx2+/– heterozygous mice in the 129sv/Bl6 background [6–8] were housed under standard laboratory conditions. A total of 31 mice were used for this study. Mice treatments and managements were approved by the Ethic Committee of the University Louis-Pasteur of Strasbourg. Colon carcinogenesis was induced in six two-month-old wild-type mice as described [7], by intraperitoneal injection of azoxymethane (AOM, Sigma, St. Louis, Missouri, USA) at 10 mg/kg once a week for five weeks. Analyses were performed 30 weeks later.

Mice were anaesthetized using isoflurane in air (5% at induction, then 1.5–2% during examination), injected subcutaneously with morphine (1 mg/kg) to inhibit intestinal motility [9]. Then, several iodine-based X-ray contrast agents were administered: Iomeron 400 (Bracco, Milan, Italy, 1.5 mL, dilution 1:5) was injected intraperitoneally; Telebrix (Guerbet, Roissy CdG, France, dilution 1:10) was administered by rectal route with a cannula until reflux from the anal orifice. In studies with the chemical carcinogen, animals were fed ad libitum for 24 h with a mixture of standard chow containing BaSO₄ powder (Micropaque, Guerbet, Roissy-CdG, France) diluted 1:10th in water to contrast the faecal matter. Dilutions of the contrast agents were chosen for differential attenuation between contrast agent inside the abdominal cavity and that inside the intestine.

Mice were placed in an isolation cell aimed at maintaining homeostatic conditions (Minerve, Esternay, France), kept anaesthetized via a mask and immobilized with a tooth bar. Imaging was performed using an X-ray μCT scanner (eXplore RS Locus, GE Healthcare, London, Canada). It produced 400 views over 360°, with three frames averaged for each view, at 80 kV, 450 μA. Data acquisition lasted about 10 min for the abdomen and about 35 min if whole-body imaging was required. After reconstruction using a Feldkamp algorithm of backprojection, the imaging volume was made of cubic voxels of 93 μm × 93 μm × 93 μm. Visualization and analysis were performed using MicroView (GE Healthcare, London, Canada).

To set up the procedure of X-ray micro-computed tomography (μCT) in living mice, we used wild-type mice and heterozygous animals for the homeobox gene Cdx2. Cdx2+/– mice exhibit gross developmental malformations in the skeleton, on the one hand, and in the gut in the form of non-cancerous polyp-like heterotopia made of gastric-type mucosa replacing the normal intestinal mucosa in the caecum region [6], on the other hand. Fig. 1A shows the apparatus for data acquisition in living mice, including the scanner for X-ray μCT and the isolation cell aimed at maintaining the mice anaesthetized and immobilized in homeostatic conditions. By μCT, calcified structures were easily observed and 3D reconstruction of the vertebral axis allowed visualizing the small homeotic bone malformations typical of Cdx2+/– mice, exemplified by anterior shifts of cervical and upper thoracic vertebrae and by supernumerary ribs (Fig. 2). Without X-ray imaging, these analyses require sacrifice of the animal and time-consuming skeletal preparations [6].

As already reported [10], the absence of natural X-ray contrast prevents any detection of soft tissues within the abdominal cavity, including the gut, without the use of exogenous contrast agent (Fig. 1Ba). Adding a single contrast agent either in the intestinal lumen (Fig. 1Bb) or in the peritoneal cavity showed the fate of the gut (Fig. 1Bc). However, even when using a single contrast agent, it is not easy to distinguish when a constriction in the gut lumen is due to a regular fold or to the presence of an abnormal structure. This problem could be overcome by demarcating both internal and external limits of the gut wall to visualize its thickness. For this purpose, we developed a multiple-contrast procedure where contrast material was administered in the peritoneal cavity and in the colonic lumen. Dilutions of contrast agents were chosen for differential attenuation between contrast agent inside the abdominal cavity and that inside the intestine, as shown by whole-body imaging (Fig. 1C). At higher magnification, this procedure allowed visualizing the colonic wall as a continuous dark line with regular thickness of about 200 μm, delineating the inside of the colonic lumen from the outside peritoneal cavity (Fig. 1Bd). The thickness is in line with previous estimations [11]. Payer's patches are small and local thickenings of the gut wall, typically located near the blunt end of the caecum and in the distal colon. By μCT, they appeared as flat elevated structures compared to the adjacent gut wall, indicating that thickenings as small as 400 μm (about twice the measurement of the normal gut wall) can be unambiguously detected by this approach (Figs. 1Bd and 3Aa, b).

Taking advantage of the gastric-type heterotopia occurring in the caecum region of Cdx2+/– mice [6], we tested whether multiple-contrast μCT is appropriate to detect these morphologically abnormal structures in the soft intestinal tissue. As shown in Fig. 3Aa and b, these lesions appeared as dark grey polyp-like structures protruding into the caecal lumen in each Cdx2+/– mouse (n=9). Subsequent macroscopic examination at necropsy demonstrated that every lesion visible on gross inspection was identified by μCT (Fig. 3Ac). Histological analysis confirmed the identity of the heterotopia (Fig. 3Ad).

Having demonstrated that multiple-contrast μCT is appropriate to detect genetically determined polyp-like lesions in the soft intestinal tissue, we set out to assess its utility in the investigation of colonic tumours. Wild-type mice received intraperitoneal injections of the carcinogen AOM for five weeks, leading to DNA methylation, point mutations and ultimately colon adenocarcinomas 30 weeks later [7]. Since a number of studies on CRC are focused on the role of nutrients, pre- and/or pro-biotics, we wished to develop a protocol that avoided starvation or dietary changes and thus did not interfere with the experimental regimen. For this purpose, mice were prepared as described above, except that starvation was omitted and replaced by ad libitum feeding with BaSO₄-containing chow to contrast the faecal matter. With this method, the stools appeared as contrasted pellets in the colonic lumen. Inversely, the endoluminal polyps induced by AOM formed dark structures connected to the gut wall (Fig. 3Ba, b, and see supplementary material for a dynamic view). A total of 11 tumours were recorded by μCT in the six mice that were analyzed, and macroscopic examination at necropsy demonstrated that none of the tumours induced by AOM escaped μCT detection; histology confirmed the presence and cancerous nature of the polyps (Fig. 2Bc and d). Thus, it is very easy to distinguish the endoluminal tumours from the faecal pellets by feeding with a contrast agent. Tumour development is a long-lasting process that takes several months in mice. Thanks to the careful control of anaesthetized mice environment and follow-up of physiological parameters, repeated anaesthesia and μCT imaging were performed up to five times on the same individual without precocious death, indicating that mouse preparation for the administration of multiple contrast agents can be recurrently performed to survey tumour growth without compromising viability. Although each X-ray examination involves potential hazard linked with ionization, the protocol was close to that described in [12], leading to moderate irradiation. Nevertheless, consequence of such a procedure during long time follow-up of wild-type and transgenic mice needs to be checked.