(A) CCR9 mRNA levels in CCR9⁺ cells of early-stage CCIC1 and CCIC2 in response to JAG1-induced NOTCH activation using quantitative PCR. HES1 was used as positive control. **P < 0.001. (B) CCR9 protein levels in CCR9⁺ early-stage CCIC1 and CCIC2 cotreated with or without proteasome inhibitor (100 nM PS341, 4 hours before harvest), in response to JAG1-induced NOTCH activation using Western blot. HES1 was used as positive control. Lanes were run on the same gel but were noncontiguous (white lines). (C) Pretreatment of CCIC with 5 μg/ml JAG1 peptide for 8 hours suppresses CCL25-dependent chemotaxis in Boyden chamber assay. *P < 0.0001, compared with control by 1-way ANOVA; **P < 0.001, compared with CCL25 alone (n = 3). Error bars indicate SEM. (D) Pretreatment of CCR9⁺ CCICs with 2 μg/ml JAG1 peptide for 8 hours suppresses CCR9 protein, and CCL25 induces phospho-AKT (Ser473) levels, with essentially no change in total AKT levels. Western blot analysis used anti-human CCR9, phospho-AKT, total AKT (AKT1, AKT2, AKT3), NICD, and HES1 antibodies. β-Actin was used as a loading control. (E) CCICs carrying GFP-NOTCH reporter were sorted into NOTCH high and low subpopulations by FACS and injected into tail veins of NOD/SCID mice. NOTCH high CCICs form more extra-GI tumors, while NOTCH low CCICs form more GI tumors. Error bars indicate SEM. **P < 0.01, *P < 0.05, compared with each other. Also see Supplemental Figures 6 and 7.