Oncogenic mutations in intestinal adenomas regulate Bim-mediated apoptosis induced by TGF-β

Significance

The TGF-β/Smad pathway is mutated in the majority of late-stage colorectal cancers (CRCs), but its role in intestinal adenomas is unclear. We show here that intestinal epithelial cells, including the Lgr5⁺ stem cells, are sensitive to the TGF-β–induced apoptosis in adenomas and that this is mediated by the BH3-only protein Bim. Furthermore, the tumor-initiating Apc mutation increases, whereas the KRas oncogene decreases the TGF-β sensitivity. Our results provide important mechanistic insight into how TGF-β regulates intestinal adenoma development and show that drugs mimicking the effects of BH3-only proteins can induce apoptosis also in CRC cells that are resistant to TGF-β.

Abstract

In the majority of microsatellite-stable colorectal cancers (CRCs), an initiating mutation occurs in the adenomatous polyposis coli (APC) or β-catenin gene, activating the β-catenin/TCF pathway. The progression of resulting adenomas is associated with oncogenic activation of KRas and inactivation of the p53 and TGF-β/Smad functions. Most established CRC cell lines contain mutations in the TGF-β/Smad pathway, but little is known about the function of TGF-β in the early phases of intestinal tumorigenesis. We used mouse and human ex vivo 3D intestinal organoid cultures and in vivo mouse models to study the effect of TGF-β on the Lgr5⁺ intestinal stem cells and their progeny in intestinal adenomas. We found that the TGF-β–induced apoptosis in Apc-mutant organoids, including the Lgr5⁺ stem cells, was mediated by up-regulation of the BH3-only proapoptotic protein Bcl-2–like protein 11 (Bim). BH3-mimetic compounds recapitulated the effect of Bim not only in the adenomas but also in human CRC organoids that had lost responsiveness to TGF-β–induced apoptosis. However, wild-type intestinal crypts were markedly less sensitive to TGF-β than Apc-mutant adenomas, whereas the KRas oncogene increased resistance to TGF-β via the activation of the Erk1/2 kinase pathway, leading to Bim down-regulation. Our studies identify Bim as a critical mediator of TGF-β–induced apoptosis in intestinal adenomas and show that the common progression mutations modify Bim levels and sensitivity to TGF-β during intestinal adenoma development.

Activation of the β-catenin/TCF transcription factor (Wnt) pathway is the initiating event in the majority of human colorectal cancers (CRCs) and one of the key regulators of CRC pathogenesis (1, 2). The β-catenin level in cells is controlled through a multiprotein complex that contains the adenomatous polyposis coli (APC) protein. In 70–80% of individuals suffering from the sporadic version of CRC, both APC alleles are mutated and thus inactivated, resulting in elevated nuclear β-catenin levels (1). Recently, the intestinal stem cells have been identified as the cells of origin of CRC (3, 4).

According to the model presented by Fearon and Vogelstein (5), the initiating APC or CTNNB1 mutation is followed by oncogenic activation of the KRAS gene and inactivation of the TGF-β signal transduction pathway and the TP53 tumor suppressor. Recent genome-wide analysis by the Cancer Genome Atlas Network has demonstrated the high frequency of mutations in the β-catenin/TCF, TGF-β, phosphoinositide 3-kinase (PI3K) and p53 pathways in CRC (6). Full understanding of the roles of the driver mutations in colorectal tumors has been hampered by lack of appropriate ex vivo model systems for studies of CRC progression, and thus it has been difficult to study the effects of TGF-β in intestinal adenoma cells at the early stages of tumor development. Although TGF-β receptors have been deleted in Apc-mutant mouse models of CRC (7, 8), these studies do not reveal the mechanism of TGF-β action in intestinal adenomas or give further insights into how the subsequent progression mutations affect the TGF-β sensitivity.

To circumvent these problems, we have here used in vivo models and the recently established intestinal ex vivo organoid cultures (9–11) for analysis of the effects of TGF-β in the early stages of intestinal carcinogenesis. In the ex vivo culture system, exogenous Wnt-ligand R-Spondin1 is essential for the survival and growth of organoids from wild-type (WT) intestinal crypts. Without R-Spondin1, only organoids with an activating mutation of the Wnt pathway survive beyond 5 d. Using the organoid cultures, we found that TGF-β exerts its apoptotic effect via the induction of the Bcl-2–like protein 11 (Bim) in early intestinal adenomas, resulting in the oligomerization of the proapoptotic protein Bak, but not Bax. We show that TGF-β induces apoptosis in vivo preferentially in the adenoma cells, including the Lgr5⁺ stem cells (12, 13), and that different stages of intestinal adenoma progression display different sensitivities to TGF-β–induced cell death.

Results

TGF-β Induces Apoptosis of Adenoma Cells via the Induction of Bim.

To study the effect of TGF-β in the early intestinal adenomas, we first cultured crypts from Apc^flox/flox; villin-Cre^ERT (ApcV) mice and induced intestinal epithelial-specific deletion of the Apc gene by 4-hydroxy-tamoxifen (4-OH-Tam) treatment (ApcV∆/∆). Similarly to organoids isolated from the Apc^min/+ mouse intestinal adenoma model (14), the ApcV∆/∆ cultures without R-Spondin1 produced spheroids, morphologically resembling adenomas, in about 8–10 d after the addition of 4-OH-Tam (Fig. S1 A–C). We checked the successful deletion of the Apc gene and efficient selection of the Apc-mutant ApcV∆/∆ organoids by genotyping, real-time PCR, and immunostaining for the β-catenin/TCF-target EphB2 (Fig. S1 D–H). In the organoids, the inner cells exfoliate via apoptosis (Fig. 1A, control), whereas TGF-β treatment induced caspase-3–dependent apoptosis also in the outer layer of the organoids, resulting in the loss of the regular smooth organoid surface (Fig. 1 A, TGF-β, dotted line, and B) and leading later to a complete disintegration (“death”) of the organoid structure (Fig. 1C). Besides TGF-β, Activin A and B, two other members of the TGF-β family, promoted organoid death in a dose- and time-dependent manner (Fig. 1D and Fig. S2A). The TGF-β–induced death was blocked by the SB431542 TGF-β receptor type-1 (TGF-βRI) kinase inhibitor and by the caspase inhibitor Z-VAD-FMK (Fig. 1 E and F), confirming the signaling pathway specificity and the involvement of active caspases in this process. However, cell death was not completely eliminated, suggesting that caspase-independent pathways are also involved. Colonic and small intestinal organoids from Apc^min/+, ApcV∆/∆, and Apc^1638N/+ (ApcN) mice, carrying another mutant Apc allele, responded similarly, indicating that the source of the mutant organoids does not influence their responsiveness to TGF-β (Fig. 1G).

Fig. 1.

Time-, dose-, and TGF-βRI kinase-dependent cytotoxicity of TGF-β in organoids with a mutant β-catenin/TCF pathway. (A) Immunostaining for active caspase-3 in ApcV∆/∆ organoids with or without TGF-β treatment (3 ng/mL for 16 h). Apoptotic cells in the outer epithelial layer of TGF-β–treated organoids are indicated with the dotted line. (B) Representative images and fold increase of active caspase-3 positivity in the TGF-β–treated ApcV∆/∆ organoids (flow cytometry measurements from three control or TGF-β–treated organoids). (C) Phase-contrast microscopic images of ApcV∆/∆ organoids with and without 48 h TGF-β treatment. (D) TGF-β dose–response curve in ApcV∆/∆ organoid cultures treated continuously for 48 or for 4 h, with analysis 44 h thereafter. (E) Effect of the TGF-βRI kinase inhibitor SB431542 (10 µM) in the cultures treated for 48 h. (F) The effect of the caspase inhibitor Z-VAD-FMK (50 μM) on the TGF-β–induced (1 h) death of ApcVΔ/Δ organoids, analyzed at 48 h. (G) Response of the indicated Apc-mutant organoids to a 4-h TGF-β treatment. The organoids isolated from the small intestine are labeled as “si” and those from colon as “co.” [Scale bars: 50 µm (A) or 100 µm (C).]

TGF-β employs a wide variety of target genes and mediators to exert its effects, depending on the cellular context (15). To elucidate the mechanism of TGF-β–regulated cell death in early intestinal adenomas, we screened a large number of proapoptotic and antiapoptotic mediators by apoptosis protein arrays and quantitative real-time PCR in the ApcV∆/∆ organoids. We found that TGF-β suppressed the expression of the Wnt-target antiapoptotic protein Birc5/survivin, stimulated the phosphorylation of Smad2 transcription factor and the MAP kinase p38, and elevated the level of the proapoptotic BH3-only protein Bim_EL isoform (Bim) (Fig. 2A and Fig. S2 B and C). In contrast, we observed only a modest increase in the mRNA level of the proapoptotic Bid and no change in Puma after TGF-β treatment (Fig. S2C). Importantly, transduction of the ApcV∆/∆ organoids with either of two shBim lentiviruses resulted in marked resistance against the TGF-β–induced death (Fig. 2 B and C).

Fig. 2.

TGF-β induces apoptosis in intestinal organoids via Bim. (A) Immunoblotting of the indicated proteins in ApcV∆/∆ organoids. (B) Effect of Bim silencing on the death of ApcV∆/∆ organoids treated with TGF-β for 1 h, analyzed at 36 h. (C) Bim immunoblotting of Scr or shBim lentivirus-transduced ApcV∆/∆ organoids. (D) Immunoblotting of Bak and Bax from gel filtrated fractions of ApcV∆/∆ organoids, cultured in the absence or presence of TGF-β for 16 h. Note that TGF-β results in the oligomerization of Bak, but not Bax (asterisks). (E) Relative levels of RNAs encoding antiapoptotic proteins in ApcV∆/∆ organoids. (F and G) Percentage of dead ApcV∆/∆ organoids and pro- and active caspase-3 immunoblotting 48 h after ABT-263 or ABT-737 treatment. (H) Representative images and quantification of flow cytometry results from untreated, ABT-263– or ABT-737–treated (1 μM for 48 h) ApcV∆/∆ organoids. (I) Immunoblotting of the indicated proteins in human CRC organoid samples derived from four patients (Pt1–4), cultured with or without TGF-β (3 ng/mL) for 48 h. (J) Immunoblotting of active CASPASE-3 in human CRC organoids 48 h after ABT-263 (10 μM) (+) or DMSO (−) treatment. (K) Percentage of active CASPASE-3⁺ cells in the human CRC organoids after ABT-263 or TGF-β treatment (flow cytometry).

Bim binds to antiapoptotic proteins, such as Bcl2, Bclx, and Mcl1, and thereby releases and activates Bcl-associated proteins Bax and/or Bak (16). The activation of Bax and/or Bak results in their oligomerization and pore formation in the mitochondrial outer membrane, leading to the release of cytochrome c into the cytoplasm, the activation of the caspase cascade, and finally to the induction of apoptosis. We observed oligomerization of Bak, but not Bax after TGF-β addition (Fig. 2D). Furthermore, we detected an abundant RNA expression of the prosurvival genes Bclx, Bcl2, and Bclw, but not Mcl1, in the Apc-mutant ApcV∆/∆ organoids (Fig. 2E). To mimic the effect of Bim, we next applied the BH3 mimetics ABT-263 and ABT-737, which are known to bind to Bclw, Bclx and Bcl2, but not Mcl1, thus resulting in the release of Bax and Bak, and subsequent apoptosis (17). Both BH3 mimetic compounds increased the proportion of dead organoids, the level of active caspase-3, and the proportion of apoptotic cells with active caspase-3 (Fig. 2 F–H). Collectively, these results indicate that TGF-β exerts its apoptotic effect primarily via Bim in Apc-mutant organoids that represent early intestinal adenomas.

To investigate the role of BIM and BH3-only proteins in CRC, we established organoid cultures from late-stage CRC patients in which the TGF-β/SMAD4 pathway is inactivated. Although TGF-β treatment did not result in elevated expression of the different BIM isoforms (BIM_EL, BIM_L, BIM_S) or apoptosis in these cultures (Fig. 2 I and K), the BH3-mimetic ABT-263 induced caspase-3 activation and organoid death (Fig. 2 J and K). Thus, the effect of BIM can be recapitulated by using BH3 mimetics in clinical samples from late-stage CRC.

TGF-β Induces Apoptosis in the Lgr5⁺ Stem Cells.

The β-catenin/TCF pathway is known to regulate intestinal stem cell markers (18). Gene set enrichment analysis of the RNA microarray data (Table S1) from ApcV∆/∆ organoids revealed that many of the β-catenin/TCF targets were suppressed in the TGF-β–treated samples compared with the controls (Fig. S3A). In contrast, no significant enrichment was detected for genes of the Notch pathway (Fig. S3A), known to function in the homeostasis of the WT intestine and in intestinal tumorigenesis (19, 20). Further Ingenuity Pathway Analysis of the microarray data indicated that TGF-β down-regulated also intestinal stem cell-specific targets, such as olfactomedin 4 (Olfm4) and tumor necrosis factor receptor superfamily member 19 (Tnfrsf19) (18) in the ApcVΔ/Δ organoids (Fig. S3B). After 18 h of TGF-β treatment, long before any morphological signs for death in the organoids, the stem cell signature genes (Lgr5, Olfm4, Tnfrsf19) showed a more profound transcriptional suppression than some other β-catenin/TCF targets, such as Axin2 or c-Myc (Fig. 3A).

Fig. 3.

TGF-β induces apoptosis of the Lgr5⁺ adenoma stem cells. (A) Effect of 18-h treatment with TGF-β on the RNA expression of selected β-catenin/TCF target genes in ApcV∆/∆ organoids. (B) Organoid initiation frequency of ApcV∆/∆ cultures in the absence or presence of 3 ng/mL TGF-β. (C) The percentage of surviving ApcL∆/∆ organoids, cultured in the absence of R-Spondin1. TGF-β was added for 48 h 4 d after 4-OH-Tam and 2 d after subculturing. (D) Active caspase-3 staining in ApcL∆/∆ organoids (11 d after the addition of 4-OH-Tam) incubated in the presence or absence of TGF-β for the indicated periods. The arrowheads mark organoid regions enriched for apoptotic Lgr5⁺ cells. (Scale bar: 50 µm.) (E) Immunoblotting of control and TGF-β–treated ApcL∆/∆ organoids for the indicated proteins. (F) Percentage of surviving ApcL∆/∆ organoids 2 and 8 d after the addition of ABT-263 (1 μM) and 4-OH-Tam, respectively. (G and H) The relative percentage of Lgr5-EGFP–positive cells and representative images of the flow cytometric analysis of control and ABT-263–treated ApcL∆/∆ organoid cultures. The data were analyzed by either paired or unpaired t test (A–G).

Although several drugs used for the treatment of cancer patients effectively kill tumor cells, many patients show disease relapse after the treatment, which is often attributed to relative drug resistance of cancer stem cells. The fact that TGF-β treatment down-regulated the intestinal stem cell-specific genes and prevented the formation of new organoids in ApcVΔ/Δ cultures (Fig. 3B) raised the possibility that the stem cells are highly sensitive to TGF-β–induced cell death. To test this hypothesis, we derived organoids from Lgr5-EGFP-IRES-Cre^ERT; Apc^flox/flox mice (ApcL) (21), which develop intestinal stem cell-derived adenomas containing Lgr5-EGFP–positive stem cells when the Cre allele is activated. Five days after the removal of R-Spondin1, the 4-OH-Tam–treated ApcL cultures contained surviving organoids (Fig. S4A), suggesting that they had undergone the Apc deletion (ApcL∆/∆). Importantly, more than 85% of the surviving Apc-mutant organoids showed Lgr5-EGFP signal (Fig. S4 B and C), specific for stem cells that contribute to the growth of established intestinal adenomas (13). TGF-β induced death in the ApcL∆/∆ organoids in a similar manner as in the ApcV∆/∆ organoids (Fig. 3C). Active caspase-3 was detected in the Lgr5-positive cells 8 h after the addition of TGF-β (Fig. 3D, arrowheads). Within 20 h, the apoptosis spread to a large portion of the organoids, resulting in decreased Lgr5 signals (Fig. 3D). TGF-β activated the p38 and Smad signaling pathways and elevated Bim expression in the organoids (Fig. 3E and Fig. S4D). Also the BH3-mimetic ABT-263 reduced the survival of the ApcL∆/∆ organoids (Fig. 3F) and decreased the size of the Lgr5-EGFP⁺ population (Fig. 3 G and H). These results indicate that TGF-β is a powerful promoter of apoptosis also in the Lgr5⁺ stem cells by inducing Bim expression.

The Apc Mutation Increases Sensitivity to TGF-β–Induced Cell Death.

To analyze how the tumor-initiating Apc mutation affects the TGF-β–induced cell death, we compared WT (ApcV) and Apc-deleted (ApcV∆/∆) crypt cultures. Unexpectedly, the ApcV cultures were more resistant to TGF-β than ApcV∆/∆ organoids under the same culture conditions (Fig. S5 A and B). The Bim level was lower in both untreated and TGF-β–treated ApcV organoids than in the ApcV∆/∆ organoids (Fig. S5C), and the ApcV organoids were less sensitive to apoptosis induced by BH3 mimetics (Fig. S5 D and E). Interestingly, the addition of either of two different insulin-like growth factor I (IGF-I) receptor inhibitors together with TGF-β increased the TGF-β–induced death in the WT cultures, but not in the Apc-deleted organoids (Fig. S5 F and G). In contrast, no effect was obtained with hepatocyte growth factor receptor inhibitor (PHA665752, 35.4 ± 4.9%, vs. solvent, 28.5 ± 6.5% dead organoids; mean ± SD; P = 0.14) or fibroblast growth factor receptor inhibitor (PD173074, 30.4 ± 10.2%, vs. solvent, 23.1 ± 11.4%; mean ± SD; P = 0.69). Taken together, these results indicate that cultures of WT crypts are more resistant to TGF-β–induced apoptosis than cultures of Apc-mutant adenomas, and that the IGF-I receptor pathway activity may partly explain their better survival.

Mutant KRas Oncogene Protects Apc-Mutant Organoids from TGF-β–Induced Death.

According to the Vogelstein model, the initiating APC or CTNNB1 mutation is often followed by oncogenic activation of the KRAS gene (5). The number of visible tumors is highly increased when the mutant KRas gene is introduced to the Apc^1638N/+ (ApcN) mice that otherwise have only few intestinal tumors (22). We cultured organoids from the intestines of ApcN and ApcN-KRas mice in conditions in which only the Apc-mutant organoids survive. A similar number of crypts gave rise to a fivefold higher number of ApcN-KRas organoids than ApcN organoids 6 d after the intestinal crypt isolation (Fig. 4A). This suggests a higher tumor initiation frequency and/or better tumor cell survival in the ApcN-KRas mice.

Fig. 4.

The KRas oncogene promotes TGF-β resistance via P-Erk1/2 in Apc-mutant organoids. (A) The effect of the KRas oncogene on the number of ApcN-mutant organoids selected without R-Spondin1 from the same number of intestinal crypts. (B) The proportion of disintegrated ApcN and ApcN-KRas organoids as a function of concentration and time of TGF-β treatment (4-h stimulation with different doses or 3 ng/mL for different durations), analyzed at 48 h (n = 3). (C) Bim and P-p38 immunoblotting of ApcN and ApcN-KRas organoids incubated with or without TGF-β for 16 h. (D) The effect of BH3-mimetics (1 µM) on active caspase-3 in ApcN-KRas organoids, detected by immunoblotting (Left) or flow cytometry (Right). (E) Analysis of phosphoproteins by antibody arrays, highlighting the P-Erk1/2 signal. (F) Immunoblotting for P-Erk1/2 and Erk1/2. (G) Confocal images of P-Erk1/2 immunofluorescence in ApcN-KRas and ApcN organoids. (Scale bar: 50 µm.) (H) Enhancement of the TGF-β (4 h)-induced death of ApcN-KRas small intestinal (si) and colonic (co), but not of ApcN intestinal organoids by the MEK1/2 inhibitor PD98059 (20 µM). Analysis was done at 48 h. (I) The effect of the MEK1/2 inhibitors PD98059 (20 µM, Inh1) and U0126 (20 µM, Inh2) on the Bim level in ApcN-KRas organoids (TGF-β for 1 h and analysis at 16 h).

Although the ApcV∆/∆ and ApcN cultures did not differ in their sensitivity to TGF-β (Fig. 1G), the ApcN-KRas organoids showed a delayed induction of apoptosis compared with ApcN after TGF-β treatment, as well as reduced Bim expression (Fig. 4 B and C, Fig. S6A). The ApcN-KRas organoid cultures showed also reduced p38 and Smad2 phosphorylation (Fig. 4C and Fig. S6B) and increased expression of Smad7, a negative regulator of the TGF-β signal transduction pathway (23) (Fig. S6B). Importantly, however, the ApcN-KRas organoids maintained their sensitivity to the BH3-mimetics ABT-263 and ABT-737 (Fig. 4D).

To study the signaling pathways downstream of KRas, we investigated the phosphorylation of the Akt protein kinase and the extracellular signal-regulated kinase (Erk) 1/2 in ApcN and ApcN-KRas organoids. As expected, the KRas oncogene increased the activating phosphorylation of Akt (Fig. S6B), but inhibitors of the upstream phosphoinositide-3 kinase (PI3K) did not abrogate the KRas-induced resistance (Fig. S6C). The ApcN-KRas organoids were also strongly and uniformly positive for P-Erk1/2, whereas the ApcN cultures displayed a much weaker signal (Fig. 4 E–G). To analyze whether P-Erk1/2 was required for the protection from TGF-β–induced death by the KRas oncogene, two different inhibitors for the upstream kinase mitogen-activated protein kinase kinase (MEK) 1/2 were applied. Both enhanced the sensitivity to TGF-β–induced cell death in ApcN-KRas, but not in the ApcN organoid cultures (Fig. 4H and Fig. S6D). Previous studies have demonstrated that Erk1/2 activation reduces Bim protein levels by Bim phosphorylation, followed by proteosomal degradation (24). Interestingly, both Erk1/2 inhibitors stabilized the basal Bim levels and restored the TGF-β–induced expression of Bim in the ApcN-KRas organoids (Fig. 4I).

To further assess the significance of our findings, we cultured organoids from tumor tissues of colorectal adenoma and carcinoma patients (SI Materials and Methods) in the absence of R-Spondin1 to ensure that only organoids with a continuously active Wnt pathway survive. Organoids derived from KRAS mutant tumors were substantially more resistant to TGF-β–induced death than organoids derived from tumors lacking KRAS mutations, reflecting our findings in the mouse model system (Fig. 5A). Furthermore, the KRAS resistance was reduced by ERK1/2 phosphorylation inhibitors, whereas the MEK1/2 inhibitor had no effect on the TGF-β–induced death in the absence of the KRAS oncogene (Fig. 5B). Taken together, our data indicate that the KRas oncogene partially inhibits the apoptosis induced by low doses of TGF-β in the CRC adenoma phase by activating the Erk1/2 pathway and by inhibiting the expression of Bim.

Fig. 5.

The KRAS oncogene promotes TGF-β resistance via P-ERK1/2 in human CRC organoids. (A) Comparison of TGF-β–induced death in organoids derived from two KRAS mutant and two KRAS WT human adenomas (Pt5–7) and a carcinoma (Pt8). (B) Effect of MEK1/2 kinase inhibitor PD98059 on the sensitivity of KRAS mutant human organoids to TGF-β–induced death (3 ng/mL for 5 h, analyzed at 72 h).

Apc and KRas Mutations Modulate the TGF-β Sensitivity of Adenoma Cells in Vivo.

To confirm the in vivo significance of our findings, we cloned the active form of mouse TGF-β1 (TGF^ACT) (25, 26) into an adeno-associated virus type 9 (AAV9) vector. We validated the expression of the vector-encoded protein by immunoprecipitation from the supernatants of transduced 293T cells (Fig. 6A). We furthermore confirmed TGF^ACT activity by stimulation of TGF-β–responsive mink lung epithelial cells (MLEC subclone; PAI1 assay; SI Materials and Methods) (27). We then induced Apc deletion in ApcL mice with a single tamoxifen injection (ApcLΔ/Δ), and after 21 d, when the mice had already visible adenomas, we injected them i.p. with AAV9-TGF^ACT or control AAV9 encoding human serum albumin (HSA) at a dose of 2 × 10¹¹ viral particles/mouse. As shown in Fig. 6A, a TGF^ACT comigrating band was increased in TGF-β immunoprecipitates from serum of AAV9-TGF^ACT-injected mice in comparison with AAV9-HSA-injected mice at 5 d (Fig. 6A).

Fig. 6.

TGF-β sensitivity of the intestinal epithelium is regulated by the Apc and KRas mutations in vivo. (A) Immunoprecipitation analysis of active TGF-β in the supernatants of 293T cell cultures and in the serum of ApcL∆/∆ mice injected with tamoxifen and AAV9-HSA or AAV9-TGF^ACT. (B and C) Prox1 and active caspase-3 immunostaining of ApcL∆/∆ intestines, and quantification in C. (D) Quantification of the active caspase-3⁺ apoptotic Lgr5⁺ cells in ApcL∆/∆ intestines. For the quantification in C and D, 20–25 adenomatous or WT crypts were selected from four mice. (E) Quantification of 20 adenoma regions from ApcN and ApcN-KRas mice 5 d after injection with AAV9-HSA or AAV9-TGF^ACT. Kruskal–Wallis or Mann–Whitney tests (C–E). (Scale bar: 50 µm.)

Analysis of active caspase-3 immunostaining indicated that at 5 d after the injection, TGF^ACT significantly elevated the proportion of apoptotic cells in the adenomas, which were marked by Prox1 immunostaining (28), whereas it had no significant effect on the WT intestinal crypts (Fig. 6 B and C). We also observed a marked increase in the proportion of apoptotic Lgr5⁺ adenoma cells in the presence of TGF^ACT (Fig. 6D). Furthermore, TGF-β was found to enhance apoptosis only in the ApcN, but not in the ApcN-KRas adenomas (Fig. 6E). Collectively, these data indicate that both WT crypts and adenoma cells carrying the KRas mutation are more resistant to the TGF-β–induced cell death than Apc-mutant adenomas and that TGF-β targets also the Lgr5⁺ stem cells in vivo.

Discussion

We show here, using mouse and human ex vivo 3D intestinal organoid cultures and transgenic in vivo models, that TGF-β induces apoptosis via the induction of the BH3-only protein Bim in intestinal adenoma cells, including the Lgr5⁺ stem cells. Our results indicate that the tumor-initiating Apc mutation increases the sensitivity to TGF-β–induced cell death, whereas the KRas progression mutation increases resistance to cell death both ex vivo and in vivo. This is consistent with the fact that stromal cell-derived TGF-β promotes CRC progression only at later stages of the disease via a mesenchyme-dependent mechanism (29). Our results further support the view that oncogenic progression gradually converts proapoptotic TGF-β signals to mechanisms that fuel tumor growth and metastasis (30). Indeed, essential elements of the TGF-β/BMP/SMAD pathway are mutated in CRC patients, indicating a selection pressure for the genetic inactivation of this signaling system in the tumor epithelium (1, 31). Consistent with this, genetic inactivation of Smad4 in an Apc-mutant background has been shown to enhance tumorigenesis and to promote a more malignant tumor phenotype (32).

Cellular responses to TGF-β are cell type and context dependent, and TGF-β can use a wide variety of pathways for apoptosis (15, 33). Our results identify Bim, a proapoptotic BH3-only member of the mitochondrial Bcl2 family of proteins, as a critical component of TGF-β–induced apoptosis in intestinal adenoma cells. Apoptosis of tumor cells is sensitive to the changes in Bim protein levels, and various tyrosine kinase inhibitors (TKIs) are known to mediate apoptosis by up-regulation of Bim (34–36). Furthermore, low Bim levels have been associated with TKI resistance (37, 38). Drugs mimicking the effects of the BH3-only proteins have been shown to induce apoptosis in different cancer models (39). We show here that the Apc-mutant organoids, modeling the early intestinal adenoma, are highly sensitive to BH3 mimetics.

Janssen et al. (22) have reported a lower rate of apoptosis in intestinal tumors of ApcN-KRas double-mutant mice than in tumors of ApcN mice. Our study provides a possible explanation for these observations as we found that the KRas oncogene increases the resistance of Apc-mutant organoids against TGF-β–induced death. However, our data show that this protective effect is only partial and is overridden by high doses or long duration of TGF-β treatment. We provide evidence that the KRas oncogene increases the activation of Erk1/2 MAP kinases, thus decreasing the levels of Bim in the Apc-mutant organoids. However, despite their resistance to TGF-β–induced apoptosis, the KRas mutant organoids were still sensitive to BH3 mimetics, a finding that could have clinical significance.

Interestingly, WT intestinal organoids were more resistant to TGF-β than the Apc-deleted organoids, but IGF-I receptor inhibitors dramatically decreased the WT resistance. IGF-I receptor signaling may provide a survival signal in the cultures due to the high concentrations of insulin (40), which is a component of the organoid culture medium. In adult humans, the IGF-I levels can reach 200 ng/mL in the blood circulation (41); thus, WT crypts could counteract the TGF-β apoptotic effects via the IGF-I receptor survival pathway not only ex vivo, but also in vivo.

Schepers et al. (13) have identified the Lgr5⁺ cells as a cell population with stem cell properties in intestinal adenomas. Importantly, this population dramatically expands after Apc deletion and produces differentiating progenies (13). Furthermore, recent results suggest that Lgr5 also marks a stem cell population with clonogenic potential in human CRC (42, 43). Although additional stem cell populations marked with lumen-forming ability exist in human CRC (44, 45), the Lgr5⁺ cells critically contribute to the initiation and growth of colorectal adenomas. Interestingly, we found that TGF-β down-regulates Lgr5 and the intestinal stem cell signature in the Apc-mutant organoid cultures. This indicates that TGF-β targets also adenoma stem cells, which fuel tumorigenesis and form the primary target of cancer therapies. The finding that TGF-β can induce apoptosis in CRC stem cells is of particular interest because of the innate resistance of cancer stem cells to many cytotoxic agents used in the clinics (46).

In conclusion, here we show that active TGF-β promotes apoptosis in intestinal adenomas, including the Lgr5⁺ stem cells, via the proapoptotic BH3-only protein Bim. Importantly, compounds mimicking the effects of Bim can recapitulate the effects of TGF-β even in TGF-β–resistant CRC patient samples. Furthermore, the WT intestinal and tumor epithelium expressing an activated, mutant KRas oncogene are more resistant to the TGF-β–induced cell death than Apc-mutant adenomas. Our data contribute to a better understanding of the complexity of TGF-β signaling in CRC development and provide clues for more effective treatment of this common disease.

Materials and Methods

A detailed description of all materials and methods can be found in SI Materials and Methods.

Intestinal Organoid Cultures.

Intestinal crypts from Apc^flox/flox; villin-Cre^ERT and Apc^flox/flox; Lgr5-EGFP-IRES-Cre^ERT mice (21) were isolated and cultured as described (9, 10, 47). Briefly, the crypts were embedded in Matrigel (BD Biosciences) and cultured in medium containing 100 ng/mL noggin, 50 ng/mL EGF (PeproTech), and 500 ng/mL mouse R-Spondin1 (R&D Systems). The crypts were subcultured into new Matrigel every 5–7 d. To activate the endogenous β-catenin/TCF pathway in the organoids, the cultures were treated with 300 nM 4-OH-Tam for 48 h. Organoids with the endogenously active β-catenin/TCF pathway were then selected and cultured in growth factor-deficient medium. Complete medium was used for comparison of the sensitivity of the WT and Apc-mutant organoids to TGF-β. TGF-β was applied at 3 ng/mL, unless otherwise indicated.

Intestinal or colonic crypts from Apc^min/+, Apc^1638N/+ (ApcN) and Apc^1638N/+; villin-KRas^V12G (ApcN-KRas) mice (22) were initially cultured without R-Spondin1. Under these conditions, only organoids with endogenously activated β-catenin/TCF pathway survived beyond 5 d.

Statistical Analysis.

Values are indicated as mean + SD (n = 3–6), unless otherwise indicated. Statistical comparison of two groups was done by two-tailed unpaired t test. Multiple groups were compared by one-way analysis of variance followed by Tukey’s post hoc test using the SPSS software. The statistical significance is marked by *P < 0.05, **P < 0.01, and ***P < 0.005.