Spontaneous colon tumor mouse strains offer numerous advantages in modeling disease. However, the wide temporal window in which lesions form and the stochastic nature of lesion location require larger cohorts for assessment of disease modulation. Reliable, reproducible and inexpensive mouse models of early-stage and invasive cancer would add to existing transgenic models. We show a new method for the creation of orthotopic murine tumors centered in the mucosal and submucosal layers anywhere in the colon, allowing creation of lesions of known age, location and extent. The system overcomes the disadvantages of heterotopic implantation and allows evaluation of lesions distally in the colon as well as proximally, thereby providing an additional method to study the effects of regionality. Invasion, host vascularization and application to disparate cell lines are demonstrated. Noninvasive imaging with magnetic resonance and colonoscopy, allowed in part by the tumor location, show potential applications of this approach. ' 2005 Wiley-Liss, Inc.Key words: colon cancer; mouse model; colonoscopy; magnetic resonance imaging A number of spontaneous colorectal tumor mouse strains have been developed that model human cancer syndromes, such as HNPCC, 1 and familial adenomatous polyposis. 2-5 These mouse models, which allow monitoring of the phenotypic effects of known genetic alterations, have proved invaluable in understanding disease course and evaluating novel therapies. 6 The typically slower growth rates in the resulting spontaneous tumors compared to implanted cell lines more closely emulate the rates of human tumor growth. However, many of the most-used mouse models form adenomas primarily in the small bowel, with relatively few lesions in the descending and sigmoid colon. Noninvasive serial imaging of small bowel lesions to evaluate anatomic size changes is quite challenging in humans and markedly more difficult in mice. Additionally, few transgenic models of invasive colorectal cancer exist; 7 the commonly used Apc Min1/2 mice typically develop adenomas but not invasive adenocarcinomas.A disadvantage for the use of these models to evaluate therapeutics is that the broad temporal variability in lesion formation and the difficulty in nondestructive serial evaluation require very large cohorts of animals to be examined over extended periods of time to realize efficacy of treatment, with mortality often used as an end point. Alternatively, heterotopic s.c. implantation of tumor cell lines suffers from incongruent host stroma, which has been shown to change tumorigenicity 8 and cell cycle regulation 9 for implanted tumors compared to implantation in the native tissue bed. Additionally, direct injection of tumor cells into murine colonic mucosa or submucosa is technically difficult and has not been reported. We hypothesized that orthotopic implantation could be achieved with focal enzymatic disruption of the colonic mucosal layer only and placement of tumor cells intraluminally. The resulting tumors ''arise'' from the muco...