Distinct Roles of HES1 in Normal Stem Cells and Tumor Stem-like Cells of the Intestine Free

Tumor stem cells (TSC), capable of self-renewing and giving rise to progeny cells in the tumor, have attracted attention as promising targets for antitumor therapeutics (1–3). However, TSC-targeting therapy may disrupt normal tissue homeostasis because most of the TSC markers are also expressed by normal stem cells (NSC; refs. 4–8). To address this concern, two strategies can be employed: the identification of TSC-specific markers (9) and the identification of functional genes that act differently in NSCs and TSCs. Testing the feasibility of the latter approach, by targeting both NSCs and TSCs, requires proper in vivo models.

Hes1 is a basic helix-loop-helix transcription factor and represses transcription by active and passive repression mechanisms (10). Because Hes1 protein negatively regulates its expression by binding to its own promoter, Hes1 expression oscillates in many cells, which plays an important role in the self-renewal of stem cells (11). Although Notch signaling is important in the maintenance of intestinal stem cells (ISC; refs. 12, 13), the downstream target required for ISC self-renewal remains elusive. Previously, we showed that the maintenance of ISCs is not affected after the prenatal triple deletion of Hes1, Hes3, and Hes5 (14). However, this result is not directly applicable to adult ISCs. Because early fetal and adult ISCs are regulated by different mechanisms (15, 16), an unknown compensation may exist in the developmental phase. The role of Hes1 in adult ISCs needs clarification to explore the possibility of Hes1-targeting tumor therapy and to provide further insight into stem cell biology.

To address these issues, we examined the role of Hes1 in the NSCs of small intestine in adult mice. Hes1 deletion in NSCs led to the loss of self-renewal capacity but did not disrupt homeostasis. We investigated the role of Hes1 in intestinal tumors. Notably, we found that Hes1 deletion in the TSCs of established intestinal tumors led to tumor regression by inducing immediate apoptosis. We propose that Hes1 may be a novel tumor-specific therapeutic target that does not perturb intestinal homeostasis.

Lgr5–EGFP-IRES-CreERT2 mice (JAX strain 008875), Bmi1-CreER mice (JAX strain 010531), Apc^Min mice (JAX strain 002020), and Rosa26-LacZ mice (JAX strain 003309) were obtained from The Jackson Laboratory. Dclk1-CreERT2-IRES-EGFP mice and Ctnnb1^lox(ex3) mice were generated as described previously (9, 17). Hes1^flox mice were kindly gifted by Dr. Ryoichiro Kageyama (Institute for Virus Research, Kyoto University, Kyoto, Japan; ref. 18). Mice were maintained on a C57BL/6 background. For induction of Cre-mediated recombination, 200 μL of 20 mg/mL tamoxifen (Sigma-Aldrich) in corn oil (for a single injection), or 100 μL of 20 mg/mL tamoxifen either twice a day over 2 consecutive days or once a day over four consecutive days (for intensive induction) was intraperitoneally injected. For experiments using normal intestinal tissue, 8-week-old mice were used. For experiments using established intestinal tumors, 6-month-old mice were used. All experiments involving mice were approved by the animal research committee of Kyoto University (Kyoto, Japan) and performed in accordance with Japanese government regulations.

Surgically resected specimens were obtained from 20 colorectal cancer patients at Kyoto University Hospital. Analyses for human subjects were approved by the ethical committee of Kyoto University Hospital and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all subjects.

For histologic analyses, mouse organs were isolated and fixed overnight in 4% paraformaldehyde, embedded in paraffin, and sectioned at a thickness of 5 μm. Sections were then deparaffinized, rehydrated, and stained with hematoxylin and eosin (H&E). For immunohistochemical analyses, sections were incubated with primary antibody overnight at 4°C and washed with PBS. Washed sections were incubated with biotinylated secondary antibody for 1 hour at room temperature. Sections were then incubated with avidin biotin–peroxidase complex (Vector Laboratories), labeled with peroxidase, and colored with diaminobenzidine substrate (Dako). For immunofluorescence, sections were incubated with primary antibody overnight at 4°C and washed with PBS. Washed sections were incubated with fluorescence-conjugated secondary antibody (Invitrogen) for 1 hour at room temperature. Primary antibodies are listed in Supplementary Table S1.

Total RNA was extracted from tissues or cells using the RNeasy Micro Kit (Qiagen). Single-strand cDNA was synthesized using a Transcriptor First Strand cDNA Synthesis Kit (Roche Applied Science). qRT-PCR was performed using SYBR Green I Master (Roche Applied Science) and LightCycler 480 (Roche Applied Science). Values are expressed as arbitrary units relative to GAPDH. Primers were listed in Supplementary Table S1.

Freshly isolated small intestine was incubated in ice-cold fixative solution (PBS containing 4% paraformaldehyde, 5 mmol/L EGTA, 2 mmol/L MgCl₂, 0.2% glutaraldehyde, and 0.02% NP-40) for 1 hour at 4°C. After washing twice in PBS for 20 minutes, tissues were incubated with LacZ substrate (PBS containing 5 mmol/L K₃Fe(CN)₆, 5 mmol/L K₄Fe(CN)₆, 2 mmol/L MgCl₂, 0.02% NP-40, 0.1% sodium deoxycholate, and 1 mg/mL X-galactosidase) overnight at room temperature. After washing twice in PBS for 20 minutes, tissues were fixed overnight in 4% paraformaldehyde in PBS at 4°C. Paraformaldehyde was removed, and the stained tissues were transferred to tissue cassettes.

Conditioned medium of L-cell line secreting Wnt3a, R-spondin 3, and Noggin (L-WRN CM) was created as described previously (19). L-WRN (ATCC; CRL-3276) was generated by H. Miyoshi (2013) as described previously (14) and passaged for fewer than 6 months. Spheroid cultures were established as described previously (19, 20). In short, crypts/glands units or tumor cell aggregates were isolated and embedded in Matrigel (BD Biosciences). For spheroid culture of normal epithelium, 50% L-WRN CM supplemented with 10 μmol/L Y-27632 (Tocris Bioscience) and 10 μmol/L SB431542 (Tocris Bioscience) was added to each well. CHIR99021 (Focus Biomolecules) and valproic acid (Tocris Bioscience) were added to the medium for Lgr5-GFP^high cell culture. For spheroid culture of intestinal tumors, advanced DMEM/F-12 (Invitrogen) supplemented with 50 ng/mL mouse EGF, 100 ng/mL Noggin (all from PeproTech), penicillin/streptomycin, 10 mmol/L HEPES, GlutaMAX, and 1×B27 (all from Invitrogen) was added to each well. For in vitro induction of Cre-mediated recombination, 1 μmol/L 4-hydroxytamoxifen (4-OHT; Sigma) was added to the cultures.

For sorting experiments, isolated crypts/glands units or tumor cell aggregates were additionally dissociated with TrypLE Express (Invitrogen) for 20 minutes at 37°C. Dissociated cells were passed through a 30-μm cell strainer and sorted by FACSAria II (BD Biosciences). Single cells were gated by forward scatter pulse width and side scatter pulse width. Dead cells were excluded by labeling with 7-AAD. The data were analyzed using FlowJo software (version 10, TreeStar).

All values are presented as mean ± SEM unless otherwise stated. The two-tailed Student t test was used for statistical analysis of continuous data. The log-rank test was used for statistical analysis of Kaplan–Meier survival curves. P values of <0.05 were considered to be significant.

To confirm the pattern of Hes1 expression, we performed IHC analysis of Hes1 using samples of Apc^Min mice and human tissues with colon cancer. Consistent with previous reports (14, 21–23), in the normal intestine, Hes1 was expressed in the crypt base columnar cells and transit-amplifying (TA) cells but not in Paneth cells or cells of the villi (Supplementary Fig. S1A and S1B). Hes1 was ubiquitously expressed in murine intestinal tumors and human colon cancers (Supplementary Fig. S1A). We sorted GFP-positive cells from intestinal epithelial cells of Lgr5-EGFP-IRES-CreERT2 mice and performed qRT-PCR. FACS distinguished a population with high expression of Lgr5-GFP (Lgr5-GFP^high) and another with low expression of Lgr5-GFP (Lgr5-GFP-low), which have been reported to correspond to stem cells and their immediate TA daughters, respectively (Supplementary Fig. S1C; ref. 20). The expression of Axin2 mRNA, which reflects Wnt/β-catenin activity, was significantly higher in the Lgr5-GFP^high population; Hes1 expression was also significantly higher in the Lgr5-GFP^high population (Supplementary Fig. S1D). The expression of Notch1 and Notch2 was significantly higher in the Lgr5-GFP^high population; however, the expression of other Notch effectors, Hes3 and Hes5, was not upregulated in the Lgr5-GFP^high population (Supplementary Fig. S1E). Wnt/β-catenin and Notch signaling has been reported to regulate Hes1 expression, respectively (10, 21, 24); these data imply that both Notch and Wnt/β-catenin signaling may be important for the upregulation of Hes1 in NSCs. The high expression of Hes1 in NSCs prompted us to examine the role of Hes1 in NSCs.

To investigate the role of Hes1 in NSCs, we deleted Hes1 in Lgr5⁺ NSCs and performed lineage tracing using Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice. After administering 4 mg tamoxifen, LacZ-labeled blue cells appeared in crypt base columnar cells at day 1, proliferated in the crypts at day 3, and gave rise to progeny cells that appeared as blue stripes from the crypts to the villi at day 5 (Fig. 1A). At day 7, these blue stripes started to disappear in the crypts, and only a few blue cells were detected on the tip of the villi (Fig. 1A). Provided that Lgr5⁺ NSCs divide every 24 hours, and epithelial cells are renewed every 4 to 5 days (25), this indicates that Hes1-deleted Lgr5⁺ NSCs can give rise to progeny cells but cannot self-renew.

To further validate these results, we maximized the induction efficiency by administering 2 mg tamoxifen twice a day over 2 consecutive days. Using this intensive induction, we also evaluated the effect of Hes1 deletion in NSCs on intestinal homeostasis. First, we sorted Lgr5⁺ NSCs 2 days after the last injection and confirmed the downregulation of Hes1 expression in Hes1-deleted Lgr5 NSCs (Supplementary Fig. S2A). Surprisingly, Hes3 and Hes5 expression was not upregulated (Supplementary Fig. S2A), suggesting that, unlike the prenatal deletion of Hes1, Hes3 and Hes5 do not compensate for Hes1 deficiency in adult NSCs. Three days after the last injection, IHC revealed that the expression of Hes1 was almost lost in the LacZ-labeled crypts of Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice (Fig. 1B and C), confirming that recombination of the floxed Hes1 allele occurred efficiently. Cleaved caspase-3 was not expressed in LacZ-labeled crypts of Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice (Supplementary Fig. S2B), thereby excluding the possibility that Hes1-deleted Lgr5⁺ NSCs or their daughter TA cells were lost due to immediate apoptosis. Five days after the last injection, the proportion of LacZ⁺ crypt–villi axes in the proximal intestine did not differ between Lgr5^CreERT2/+; Hes1^+/+; Rosa26^LacZ/+ and Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice (Supplementary Fig. S2C and S2D). Fourteen days after the last injection, most of the epithelial cells in proximal small intestine, where Lgr5⁺ NSCs are abundantly detected, were labeled blue in Lgr5^CreERT2/+; Hes1^+/+; Rosa26^LacZ/+ control mice (Fig. 1D and E; Supplementary Fig. S2D). This indicates that, with this intensive induction, recombination occurred in most of the Lgr5⁺ cells in proximal small intestine. Conversely, in Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice, most of the LacZ-labeled blue cells were depleted and repopulated with LacZ⁻ cells (Fig. 1D and E; Supplementary Fig. S2D). These results further indicate that Hes1-deleted Lgr5⁺ cells cannot self-renew. The repopulated LacZ⁻ intestinal epithelium of Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice did not display obvious morphologic abnormalities (Fig. 1E). Immunohistochemical analysis revealed newly generated LacZ⁻ Lgr5⁺ cells at the base of the crypts and LacZ⁻ Hes1⁺ cells in the crypts of Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice (Fig. 1E). The number of Lgr5⁺ cells and Hes1⁺ cells did not differ between Lgr5^CreERT2/+; Hes1^+/+; Rosa26^LacZ/+ control mice and Lgr5^CreERT2/+; Hes1^flox/flox; Rosa26^LacZ/+ mice at day 14 (Fig. 1C). Although Lgr5⁺ NSCs cannot self-renew after the deletion of Hes1, Lgr5⁺ NSCs were newly generated, likely from some of the Lgr5⁻ reserve stem cells (e.g., Bmi1⁺ cells), as previously reported (26). Furthermore, we also evaluated the effect of Hes1 deletion on the proliferative capacity and secretory cell differentiation before (at day 5) and after (at day 14) replacement with LacZ⁻ cells; however, no significant difference was found (Supplementary Fig. S3A-C).

Lgr5-EGFP-IRES-CreERT2 mice have mosaic expression of the reporter (27); there remains some unlabeled Lgr5⁺ cells that escape the induction. To address this issue, we sorted single Lgr5-GFP^high cells from the normal intestinal epithelia of Lgr5^CreERT2/+; Hes1^+/+ and Lgr5^CreERT2/+; Hes1^flox/flox mice; this approach circumvents mosaic reporter expression. Then, we cultured Lgr5-GFP^high cells in Matrigel, in 50% L-WRN CM, CHIR99021, and valproic acid, which facilitates homogenous Lgr5-GFP^high stem cell culture, with 4-OHT induction for 48 hours from the start of the culture (Fig. 2A). Compared with that of the control, colony formation was almost completely abrogated in Hes1-deleted Lgr5-GFP^high single cells (Fig. 2B and C), showing that Hes1 is required for the self-renewal of Lgr5-GFP^high stem cells. We also assessed the effect of Hes1 deletion in Lgr5⁺ cells on established spheroids (Fig. 2D). Spheroids grown in 50% L-WRN CM were enriched for stem and/or progenitor cells (19). The Lgr5 expression level in the spheroids decreased when cultured without CHIR99021 and valproic acid (Fig. 2E), suggesting the presence of Lgr5⁻ TA cells in these cultures. Three days after induction, the growth of established spheroids, cultured in 50% L-WRN CM, was not affected, and Lgr5⁺ cells were detected (Fig. 2F and G); these Lgr5⁺ cells were likely generated from the Lgr5⁻ TA cells. These data demonstrate that deletion of Hes1 in NSCs does not affect intestinal homeostasis.

Bmi1 is expressed in NSCs of the intestine (6); Bmi1⁺ cells serve as reserve stem cells in cases of injury or Lgr5⁺ cell ablation (26). We assessed whether Hes1 is important in the maintenance of these NSCs that can serve as reserve stem cells. We deleted Hes1 in Bmi1⁺ NSCs and performed lineage tracing using Bmi1^CreER/+; Hes1^flox/flox; Rosa26^LacZ/+ mice. After injection with 4 mg tamoxifen, LacZ-labeled blue cells appeared four or five cells above the base of the crypt at day 1 and gave rise to progeny cells that arranged into blue stripes, reaching from the crypts to the villi, at day 3 (Fig. 3A). From days 5 to 7, blue stripes started to disappear in some crypts, and only a few blue cells were detected on the tip of the villi (Fig. 3A).

We maximized the induction efficiency by administering 2 mg tamoxifen over 4 consecutive days. IHC was used to confirm that recombination of the floxed Hes1 allele occurred efficiently (Fig. 3B and C). Ten days after the last injection, most of the LacZ-labeled blue cells were depleted and repopulated with LacZ⁻ cells in Bmi1^CreER/+; Hes1^flox/flox; Rosa26^LacZ/+ mice compared with those in Bmi1^CreER/+; Hes1^flox/+; Rosa26^LacZ/+ control mice (Fig. 3D and E). The number of Hes1⁺ cells did not differ between Bmi1^CreER/+; Hes1^flox/+; Rosa26^LacZ/+ mice and Bmi1^CreER/+; Hes1^flox/flox; Rosa26^LacZ/+ mice at day 10 (Fig. 3C and E). These data demonstrate that, regardless of the stem cell markers tested, Hes1 is necessary for the self-renewal of NSCs; however, Hes1 deletion in NSCs does not perturb intestinal homeostasis.

To investigate the role of Hes1 in tumorigenesis, we stabilized β-catenin for intestinal tumor formation and deleted Hes1 in Lgr5⁺ cells using Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice. Twenty-eight days after injection with 2 mg tamoxifen over 4 consecutive days, we found that only microadenomas were formed in Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice, whereas large tumors were formed throughout the intestine in Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^+/+ control mice (Fig. 4A). IHC revealed the expression of Hes1 and nuclear accumulation of β-catenin in the large tumors of Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^+/+ mice and microadenomas of Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice (Fig. 4A). These results suggest that microadenomas in Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice were formed only from Lgr5⁺ cells that escaped deletion of Hes1 but had not escaped β-catenin stabilization. Because of the strongly suppressed tumor formation, Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice survived significantly longer than the control mice (median survival time: 96 days vs. 27 days, respectively; Fig. 4B). We analyzed the number and size of adenomas in Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice at different times after the administration of tamoxifen. We found that the total number of adenomas did not increase, but the size of a few nascent adenomas gradually increased over time, resulting in the formation of some microadenomas and a few macroadenomas at 15 weeks (Fig. 4C). IHC revealed the expression of Hes1 in microadenomas, macroadenomas, and the normal intestinal epithelia of Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox mice (Fig. 4D), reaffirming that tumors were formed only from Lgr5⁺ cells that had escaped Hes1 deletion. We performed lineage tracing using Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^+/+; Rosa26^LacZ/+ and Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox; Rosa26^LacZ/+ mice. Twenty-eight days after induction, most of the LacZ-labeled blue cells were depleted and repopulated by LacZ⁻ epithelium in Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^flox/flox; Rosa26^LacZ/+ mice, whereas LacZ-labeled blue intestinal tumors formed in Lgr5^CreERT2/+; Ctnnb1^lox(ex3)/+; Hes1^+/+; Rosa26^LacZ/+ control mice (Fig. 4E and F). These results confirm that β-catenin stabilization cannot offset the inability of Hes1-deleted Lgr5⁺ NSCs to self-renew, which leads to substantial suppression of intestinal tumor formation.

We have shown that concomitant Hes1 deletion in β-catenin–stabilized NSCs abrogates tumor formation; however, to recapitulate Hes1-targeting tumor therapy, it is necessary to examine the effect of Hes1 deletion in established intestinal tumors. It has been shown that Lgr5 marks TSCs in the intestinal tumors of Apc-deficient mice (5). To investigate the role of Hes1 in the TSCs of established intestinal tumors, we deleted Hes1 in Lgr5⁺ TSCs of the intestinal tumors and performed lineage tracing using Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+; Rosa26^LacZ/+ mice. Five days after 4 mg tamoxifen administration, normal intestinal epithelia were intact and labeled blue; however, the intestinal tumors were injured and devoid of blue cells in Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+; Rosa26^LacZ/+ mice (Fig. 5A). In the Lgr5^CreERT2/+; Hes1^+/+; Apc^Min/+; Rosa26^LacZ/+ control mice, the normal intestinal epithelia and the intestinal tumors were labeled blue (Supplementary Fig. S4A and S4B). We maximized induction efficiency by administering 2 mg tamoxifen over 4 consecutive days. The intestinal tumors were more severely injured at day 5 (Fig. 5A); the blue cells of the normal intestinal epithelia and the injured intestinal tumors were depleted at day 10 in Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+; Rosa26^LacZ/+ mice (Fig. 5A).

To further evaluate the effect of Hes1 deletion in Lgr5⁺ TSCs, we followed the time course of LacZ-labeled blue cells in the intestinal tumors after a single 4 mg injection of tamoxifen in Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+; Rosa26^LacZ/+ mice. Blue cells in the tumors were present at day 1 and depleted by day 3 (Fig. 5B). At day 1, IHC showed cleaved caspase-3 expression in Hes1-deleted Lgr5⁺ TSCs (Fig. 5B). The proportion of cleaved caspase-3⁺ cells in Lgr5⁺ cells was significantly higher in the intestinal tumors of Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice than in those of Lgr5^CreERT2/+; Hes1^+/+; Apc^Min/+ mice (15% vs. 2.7%, respectively), whereas cleaved caspase-3 expression was not detected in Lgr5⁺ cells of the normal intestinal epithelia of Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice at day 1 (Fig. 5C). To further assess the behaviors of Lgr5⁺ NSCs and Lgr5⁺ TSCs after the deletion of Hes1, we sorted Lgr5⁺ NSCs and Lgr5⁺ TSCs from Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice at day 1 after a single 4 mg injection of tamoxifen and used qRT-PCR to compare the levels of mRNA expression with those in the control mice. Hes1 expression was downregulated in Lgr5⁺ NSCs and Lgr5⁺ TSCs at day 1 after induction (Fig. 5D; Supplementary Fig. S4C). Of note, the expression of antiapoptotic gene Bcl2 was significantly downregulated, and the expression of proapoptotic gene Bax had a tendency to be upregulated in Hes1-deleted TSCs, whereas the expression levels of Bax and Bcl2 were not altered in Hes1-deleted NSCs (Fig. 5D; Supplementary Fig. S4C). Previous studies showed that Hes1 directly suppresses Pten (28, 29), which upregulates p53 activity (30) and downregulates the PI3K–Akt pathway (31) to induce apoptosis. Interestingly, Pten was upregulated in the Hes1-deleted TSCs, but not in the Hes1-deleted NSCs at day 1 (Fig. 5D; Supplementary Fig. S4C). These results demonstrate that Hes1 deletion results in immediate apoptosis of Lgr5⁺ TSCs, but not Lgr5⁺ NSCs.

To evaluate whether Hes1 deletion in Lgr5⁺ TSCs can be used as a tumor-specific treatment, we analyzed the number of intestinal tumors in Lgr5^CreERT2/+; Hes1^+/+; Apc^Min/+ control mice and Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice after administering 2 mg tamoxifen over 4 consecutive days. Compared with Lgr5^CreERT2/+; Hes1^+/+; Apc^Min/+ mice, the number of intestinal tumors was significantly reduced in Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice 5 days after intensive induction (Fig. 5E). IHC revealed that the number of Lgr5⁺ cells was unaltered in the normal intestinal epithelia, but significantly reduced in the remaining few intestinal tumors of Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice at day 5 (Fig. 5F). These results suggest that, in contrast with the normal intestinal epithelia, no new Lgr5⁺ cells were generated after Hes1 deletion in the intestinal tumors.

To observe the time course of the same intestinal tumors after Hes1 deletion, we generated spheroids (19) from the intestinal tumors of Lgr5^CreERT2/+; Hes1^+/+; Apc^Min/+ and Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice, using a medium containing EGF and Noggin; we then compared the spheroid growth after 4-OHT administration for 24 hours (Supplementary Fig. S5A). The spheroids from Lgr5^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice ceased to expand, and Lgr5-GFP⁺ cells were depleted at day 3 (Supplementary Fig. S5B and S5C), in contrast with the unaltered growth of established spheroids generated from normal intestinal epithelia (Fig. 2D–G). These results demonstrate that Hes1 deletion in Lgr5⁺ cells specifically suppresses tumor growth.

We have recently identified Dclk1 as a TSC-specific marker in the intestine by murine lineage-tracing experiments; we proposed that Lgr5⁺Dclk1⁺ cells are the bona fide TSCs in the intestinal tumors (9). To further confirm that Hes1 deletion in TSCs leads to tumor regression, we deleted Hes1 in Dclk1⁺ TSCs of established intestinal tumors and performed lineage tracing using Dclk1^CreERT2/+; Hes1^flox/flox; Apc^Min/+; Rosa26^LacZ/+ mice. Five days after administering 4 mg tamoxifen, the intestinal tumors of Dclk1^CreERT2/+; Hes1^flox/flox; Apc^Min/+; Rosa26^LacZ/+ mice were injured and devoid of blue cells (Fig. 6A), which contrasted the blue intestinal tumors in Dclk1^CreERT2/+; Hes1^+/+; Apc^Min/+; Rosa26^LacZ/+ control mice (Fig. 6A). Next, we followed the time course of LacZ-labeled Hes1-deleted Dclk1⁺ cells after induction. Blue cells were present in the intestinal tumors at day 1 and depleted by day 3 (Fig. 6B). IHC revealed cleaved caspase-3 expression in Dclk1⁺ cells at day 1 (Fig. 6C), suggesting that deletion of Hes1 in Dclk1⁺ TSCs also induces immediate apoptosis.

We maximized induction efficiency by administering 2 mg tamoxifen over 4 consecutive days and analyzed the number of intestinal tumors in Dclk1^CreERT2/+; Hes1^+/+; Apc^Min/+ and Dclk1^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice. Compared with the number of intestinal tumors in the control mice, the number of intestinal tumors was significantly reduced in Dclk1^CreERT2/+; Hes1^flox/flox; Apc^Min/+ mice 5 days after intensive induction (Fig. 6D). These results demonstrate that, irrespective of stem cell markers analyzed, Hes1 deletion in the TSCs leads to tumor regression by inducing immediate apoptosis in the TSCs.

In this study, we focused on the specific role of Hes1 in the NSCs and TSCs of the intestine and showed that the different roles of Hes1 can enable stem cell–targeting tumor therapy. We first demonstrated that the deletion of Hes1 in the NSCs results in loss of self-renewal but does not perturb homeostasis. The Notch and Wnt/β-catenin pathways play a crucial role in the maintenance of NSCs (12, 13, 32). Hes1 is a well-known Notch effector (10), and its expression is also directly regulated by the Wnt/β-catenin pathway (21, 24). Our study showed that although the expression of other Notch effectors, Hes3 and Hes5, was not upregulated, the expression of Hes1 was upregulated in the NSCs. Although the interaction between Notch and Wnt/β-catenin pathways is complex and context dependent (13, 21, 33, 34), their role in the maintenance of NSCs may be dependent on their common downstream target, Hes1.

Previously, we employed the Villin-Cre mice to show that prenatal Hes1 deletion induces transient secretary differentiation, which is compensated by Hes3 and Hes5, and that the maintenance of NSCs is unaffected even after the prenatal triple deletion of Hes1, Hes3, and Hes5 (14). We speculate that other Notch effectors (e.g., Hey1, Hey2, and HeyL) had compensated in the maintenance of NSCs. Recombination in Villin-Cre mice occurs early in the developmental stage (35) when the intestinal epithelium is generated from progenitors other than Lgr5⁺ cells (15, 16); these progenitors may transiently permit Hes1 deficiency and provide time for the compensation. In contrast, this study showed that Hes1 deficiency was not compensated in adult NSCs, indicating a novel role of Hes1 in the maintenance of adult NSCs, which was masked in our previous analysis of prenatal Hes1 deletion. However, Lgr5-EGFP-IRES-CreERT2 mice have mosaic expression of the reporter (27); we cannot exclude the possibility that Hes1-expressing unlabeled Lgr5⁺ cells had replaced Hes1-deleted Lgr5⁺ cells by neutral drift dynamics (36) in vivo. To address this, we sorted Lgr5-GFP^high cells and showed that Hes1-deleted Lgr5⁺ stem cells cannot self-renew in vitro. Hes1 deletion using stem cell–specific reporter mice without the mosaic problem (27) could further corroborate the role of Hes1 in vivo. In contrast to prenatal Hes1 deletion, Hes1 deletion in adult NSCs did not increase the number of secretory cells (e.g., goblet cells, enteroendocrine cells, and Paneth cells). This was probably because the depletion of Hes1-deleted NSCs did not provide sufficient time for the differentiation shift.

After Hes1 deletion in Lgr5⁺ cells, normal intestinal epithelia were replaced by wild-type epithelia with normal Hes1 expression in the crypts. The intact epithelium following Hes1 deletion in Lgr5⁺ NSCs could be explained by the previous report showing that ablation of Lgr5⁺ NSCs itself does not disrupt homeostasis because of compensation by reserve stem cells (26). To address the issue, we deleted Hes1 in Bmi1⁺ NSCs, which have the potential capacity of reserve stem cells (26). It has been shown that, in contrast to the ablation of Lgr5⁺ NSCs, the ablation of Bmi1⁺ NSCs leads to crypt loss, perturbing the intestinal homeostasis (6). We showed that Hes1 deletion in Bmi1⁺ cells does not perturb homeostasis either. It is possible that adjacent TA cells compensate for the loss of reserve stem cells, because Hes1 deletion, which results in loss of self-renewal, has milder influences on the adjacent TA cells than does ablation. Further studies are required to corroborate this speculation.

Several reports have shown the potential of Hes1 as a target for antitumor therapeutics. Hes1 knockdown suppresses cell proliferation in T-cell acute lymphoblastic leukemia and colon cancer cell lines (29, 37, 38), and Hes1 overexpression induces self-renewal properties in colon cancer cell lines (23). Our previous study showed that when Hes1 is deleted prenatally, the characteristics of Apc^Min tumors mimic those of normal intestinal epithelia, such as differentiation and low proliferative capacity (14). However, it is possible that compensation in the developmental phase permitted atypical tumor formation; additionally, we did not examine apoptosis or immediate changes after postnatal Hes1 deletion. Therefore, in this study, we performed TSC-specific Hes1 deletion, which indicated that the deletion of Hes1 in the TSCs of established tumors leads to tumor regression by inducing immediate apoptosis. These findings provide strong evidence for Hes1-targeting tumor therapy. The dramatic change in Hes1-deleted Lgr5⁺ TSCs can be attributed to the abrupt upregulation of Pten. Hes1 directly suppresses Pten (28, 29), and Pten upregulation induces apoptosis via direct upregulation of p53 activity (30) and downregulation of the PI3K–Akt pathways, such as the PI3K–Akt–mTOR and PI3K–Akt–GSK3β pathways (29, 31). Provided that Hes1 knockdown in the wild-type and mutant p53 colon cancer cell lines increases apoptosis (29), multiple pathways, such as Pten-p53 and Pten-Akt, may be involved in inducing apoptosis in Hes1-deleted TSCs.

In contrast to normal intestinal epithelia, Lgr5⁺ TSCs were not generated from the surrounding tumor cells after deletion of Hes1. The loss of Lgr5⁺ TSCs after Hes1 deletion can be attributed to the following two factors. First, the deletion of Hes1 in the TSCs results in changes that are more severe than those in NSCs (immediate apoptosis in the former compared with loss of self-renewal in the latter), and so the surrounding cells could not compensate for the loss of Lgr5⁺ TSCs due to apoptosis. Second, because the population of Lgr5⁺ cells expanded in the intestinal tumors (2.9% in the normal intestine compared with 13% in the intestinal tumors; Fig. 5F), it is possible that the Lgr5⁺ cells in the intestinal tumors consist of both stem cells and progenitor cells. The first factor may be more plausible because Hes1 deletion in Dclk1⁺ TSCs also leads to tumor regression by inducing immediate apoptosis. Using murine lineage-tracing experiments, we have recently shown that Dclk1 marks TSCs in the intestine and proposed that Lgr5⁺Dclk1⁺ cells are the bona fide TSCs in intestinal tumors (9). In this study, we showed that Hes1 deletion in Dclk1⁺ cells as well as in Lgr5⁺ cells leads to tumor regression, implying that Hes1 deletion in Lgr5⁺Dclk1⁺ cells, which account for only a small subset of the tumor cells, may be sufficient for tumor regression.

Inhibition of Hes1 by small molecules, and in vivo delivery of Hes1 siRNA, are two approaches that can be employed to target Hes1 for therapeutic purposes. Several compounds can inhibit Hes1 activity (38–40). Agalloside, which directly binds to Hes1, has been recently identified as a naturally occurring Hes1 dimer inhibitor (39). Furthermore, siRNA delivery systems are being developed for cancer treatment and have already entered clinical trials (41). The advance of these technologies may realize the potential of Hes1-targeting tumor stem cell therapy in the future.

In summary, using several murine models with different NSC and TSC markers, we have shown the novel role of Hes1 in the maintenance of NSCs and TSCs in the adult intestine. Hes1 deletion in the NSCs leads to loss of self-renewal but does not disrupt homeostasis (Fig. 7A). Conversely, Hes1 deletion in the TSCs leads to tumor regression by inducing immediate apoptosis (Fig. 7B). On the basis of the different roles of Hes1 in the NSCs and TSCs, we propose that Hes1 can be used as a novel tumor-specific therapeutic target that does not perturb intestinal homeostasis.

No potential conflicts of interest were disclosed.

Conception and design: N. Goto, T. Ueo, H. Seno

Development of methodology: N. Goto, T. Ueo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): N. Goto

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Goto, M.M. Taketo

Writing, review, and/or revision of the manuscript: N. Goto, T. Ueo, A. Fukuda, M.M. Taketo, T. Chiba, H. Seno

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Kawada, H. Miyoshi, T. Chiba

Study supervision: A. Fukuda, H. Seno

The authors thank Ryoichiro Kageyama for kindly providing Hes1^flox mice and Tetsuo Sudo for kindly providing the Hes1 antibody.

This work was supported in part by Grants-in-Aid KAKENHI (25112707, 26293173, 15J03212, 16K09394, 16K15216, and 16K15427) and a research program of the Project for Development of Innovative Research on Cancer Therapeutics (P-Direct) from the Ministry of Education, Culture, Sports, Science, and Technology, and the Japan Society for the Promotion of Science. This work was also supported by Health Labour Sciences Research Grants from the Ministry of Health, Labour, and Welfare (the development of innovative therapeutic drug for the intractable inflammatory bowel disease), the Kobayashi Foundation for Cancer Research, the Naito Foundation, Princess Takamatsu Cancer Research Fund13-24514, the Mochida Foundation, and the Japanese Society of Gastroenterology.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.