Kras/ADAM17-Dependent Jag1-ICD Reverse Signaling Sustains Colorectal Cancer Progression and Chemoresistance

Sporadic colorectal cancer development is characterized by well-known histopathological changes, resulting from specific genetic defects in selected oncogenes and tumor-suppressor genes. The most of sporadic colorectal cancers and hereditary colorectal tumors show loss of APC function, the negative regulator of Wnt signaling, ultimately leading to abnormal β-catenin–dependent gene expression (1). In intestinal epithelial cells, constitutive activation of β-catenin/TCF leads to adenomatous polyp formation, a first step toward colorectal cancer development. In addition, Ras driver mutations, found in about 50% of all colorectal cancers and in advanced adenomas (2, 3), strongly sustain the pathogenesis of colorectal cancer, regulating tumor cell proliferation, survival, invasion, metastasis formation, and drug resistance (2, 4). Constitutive activation of Kras is one of the best-characterized events in colorectal cancer development, able to trigger multiple downstream pathways, including the RAF/MEK/Erk MAPK and the PI3K–AKT effector pathways (5). Several observations suggest an involvement of MEK/Erk signaling in intestinal tumorigenesis (6), but the exact molecular mechanisms remain unclear. Of note, a growing body of evidence shows that the oncogenic Kras regulates ADAM17 activity and the shedding of several growth factors in a MEK/Erk-dependent manner (4, 7). Kras mutations confer colorectal cancer resistance to anti-EGFR therapy and are associated with a worse prognosis (8). Current therapeutic options for advanced colorectal cancer have not dramatically improved clinical outcomes of patients with metastatic colorectal cancer. Therefore, a better understanding of molecular mechanisms involved in colorectal cancer development and progression is imperative for the improvement of therapeutic approaches.

Interestingly, recent studies have revealed that a sustained activation of β-catenin/TCF is responsible for transcriptional activation of Notch-ligand Jagged1, resulting in an upregulation of Jagged1 that is required for tumorigenesis in the intestine (9). High-expression levels of Jagged1 are associated with increased progression, metastatic potential, recurrence, and poor prognosis in several human malignancies, as prostate, renal, head and neck cancer and colorectal cancer (10–13). The commonly accepted scenario is based on the idea that Jagged1 ligand is able to contribute to tumorigenesis by activating canonical Notch signaling (14).

Jagged1 belongs to the Delta, Serrate, Lag-2 (DSL) family of single-pass transmembrane ligands, including Delta-like (DLL1, 3 and 4) and Jagged (Jagged 1 and 2) that transactivate the Notch receptors (Notch1–4) in signal-receiving cell (14), through a direct contact. Receptor/ligand interaction renders Notch susceptible to proteolytic processes mediated by A-Disintegrin Metalloprotease ADAM-10 and PS/γ-secretase protein complex, which ends in the release of its intracellular domain (Notch-ICD). Notch-ICD moves into the nucleus where it binds to RBP-Jκ transcription factor and recruits coactivators to form a transcription-activating complex to activate several downstream effectors, such as hairy and enhancer of split (Hes). Aberrant activation of Notch signaling is frequently observed in many human cancers (15–17), including colorectal cancer (18).

Emerging evidences indicate that Jagged1 is processed in a fashion similar to Notch by sequential proteolytic cleavages that involve two distinct enzymes: ADAM-17/TACE and PS/γ-secretase complex, ultimately resulting in the release of a nuclear-targeted intracellular domain (Jag1-ICD), that may play an important role in tumor development and carcinogenesis (19–21), possibly interacting and/or empowering the activation of other deregulated signaling pathways (22, 23).

In this article, we demonstrate that the function of Jagged1 may go beyond its effect on canonical Notch activation in colon malignancies. Indeed, we observed that in colorectal cancer cells with Kras activation, the Jagged1 ligand is not only abundantly expressed, but it undergoes a constitutive processing that ends in the aberrant generation of an intracellular fragment (Jag1-ICD), capable to move into the nucleus and to induce intrinsic reverse signaling, exerting regulatory effects on colorectal cancer tumor biology. A Kras/Erk/ADAM17 axis constitutively triggers Jag1-ICD nuclear accumulation, which favors tumor development, progression, and chemoresistance through a noncanonical mechanism.

The 6-week-old female CD1 nude mice were purchased from Charles River Laboratories Italia s.r.l. and were housed in the Institute's Animal Care Facilities.

All animal experiments were approved by local ethic authorities and conducted in accordance with Italian Governing Law (D.Lgs. n.26/2014/Protocol Number: C1368.4) and European Directive 2010/63/UE

The following human colon cell lines CCD18-Co (CRL-1459), HT29, HCT15, DLD1, HCT116, LS174T, LoVo, RKO, SW1116, and SW948 were purchased from the ATCC. Cell lines were subjected to routine cell line quality controls (e.g., morphology, Mycoplasma #G238, Abm Inc.) and authenticated by DNA profiling (short tandem repeat, STR) by the cell bank prior to shipping. The culture media were supplemented with 1% Glutamine (ECB3000D, Euroclone), 1% Antibiotics (ECB3001D, Euroclone), and 10% regular FBS (Heat-Inactivated; Life Technologies). The media were renewal 2 to 3 times per week. Cells recovered from frozen aliquots were allowed one passage to reach exponential growth phase following recovery before being used. Cells at passages greater than 10 were not used.

An opportune amount of cells was treated with different compounds: 50 μmol/L TAPI-2 (# 55123-66-5; Peptides International Inc.), 200 ng/mL of Phorbol 12-myristate 13-acetate (#P8139, Sigma-Aldrich), 30 μmol/L di U0126 (#662005, Calbiochem), with 5-fluorouracil (5FU; #F6627, Sigma-Aldrich) or irinotecan (#134760, Sigma-Aldrich).

A total of 1 × 10⁶ HCT15 cells, treated with TAPI-2 compound or vehicle alone, were fixed for 30′ in EtOH 70%, washed in PBS, treated with 100 μg/mL RNase A (cat. #R6513, Sigma-Aldrich) for 15′ and then incubated with 10 μg/mL propidium iodide (cat.#P4170) for 30′. The stained cells were analyzed on a FACS-Calibur with CellQuest software (BD Biosciences; ref. 24).

For generating cell lines stably overexpressing Jag1-ICD, murine Jag1-ICD cDNA was amplified by RT-PCR (Supplementary Table S1) and cloned into pcDNA 3.1/V5-His TOPO TA Expression Kit (#KJ48001-01, Invitrogen by Life Technologies) by following the manufacturer's instructions. V5-Jag1-ICD plasmid or pcDNA3-Neo was used to transfect HCT15 cell line using Lipofectamine 2000 (Life Technologies), according to the manufacturer's instructions. Forty-eight hours post-transfection, the cells were cultured in selection medium containing 800 ng/mL neomycin (#A1720, Sigma-Aldrich) for 4 weeks. pBABE-PURO (#1764) and pBABE K-RAS 12V (#12544) retroviral constructs were purchase from AddGene. Phoenix packaging cells were transfected with retroviral vectors by Lipofectamine 2000. After 48 hours of incubation at 32°C, the supernatants containing viral particles were collected and infection of CCD18-Co cells was performed, by using a 2 μgr/mL of Polybrene. Stable clones were obtained by using 1.5 μgr/mL for puromycin for one week.

Total RNA extraction and reverse transcription-PCR (RT-PCR) were previously described (25, 26). One μg of RNA was processed for RT-PCR using SensiFAST cDNA Synthesis Kit (Bioline). Analysis of gene expression was realized by qPCR using Taq-Man designed assays (Supplementary Table S1; Dharmacon Inc.) on the StepOnePlus Real-Time PCR System (Applied Biosystems, Life Technologies), following the manufacturer's protocol for the comparative C_t method. Data were analyzed by the ΔΔC_t method and GAPDH was used for normalization (27).

RNA silencing was performed using 100 nmol/L of Jagged1 (cat. #L-011060-00-0005) or Kras (cat. #L-005069-00-0005) ON-TARGET plus SMART pool small interference RNA (siRNA) or scrambled (cat. #D-001810-10-20, Dharmacon Inc.), using Lipofectamine RNAiMAX (Life Technologies), according to the manufacturer's instructions.

Whole-cell extract (WCE; ref. 28), extracellular shed protein preparations (29), subcellular fractioning (30), and immunoblot assay with the described antibodies (Supplementary Table S2; ref. 31) were performed as described elsewhere. Bound antibodies were detected with enhanced chemiluminescence (ECL kit, Amersham, GE Healthcare). To perform immunoprecipitation assay (32), an equal amount of WCE derived from HCT15 or DLD1 cell lines, treated with the opportune dose of phorbol 12-myristate 13-acetate (PMA) or vehicle, were precleared with Protein A-Agarose (cat. #sc-2001; Santa Cruz Biotechnology); immunoprecipitation assay was realized with ADAM17 antibody (Supplementary Table S2) or normal IgG (cat. #sc-2027; Santa Cruz Biotechnology) overnight at 4°C. The complexes were precipitated with Protein A–Agarose, and the post-transductional modifications were evaluated by using anti–phospho-serine antibody (Supplementary Table S2; ref. 33).

Tissues were fixed in 4% formalin and paraffin embedded. Consecutive sections (2-μm-thick) were stained with hematoxylin and eosin. Immunocytochemical assay was performed using an anti-Jagged1 antibody (Abcam; Supplementary Table S2). Detection was carried out with Mouse-to-Mouse HRP (DAB) staining system (ScyTek Laboratories), according to the manufacturer's instructions. Images were acquired with a Leica DM1000 microscope equipped with a ProgRes Speed XTcore 3 CCD camera and collected using ProgRes CapturePro 2.8 software (Jenoptik Optical Systems GmbH; ref. 34).

Chromatin immunoprecipitation (ChIP) was performed as described earlier (35). One μg of specific antibodies (Supplementary Table S2), or normal IgG (cat. #sc-2027, from Santa Cruz Biotechnology) was used for immunoprecipitation. In silico analysis using MatInspector (Genomatix Software GmbH, Munich, Germany) allowed us to identify predicted binding sites for RBP-Jκ on human snail1 and snail2 promoters, racing from −1,860 to −1,847 for snail1 and from −5,091 to −5,087 for snail2 (Supplementary Table S1).

HCT15 cells stably transfected with V5-Jag1-ICD–expressing vector or pCDNA3-Neo control were plated in 96-well plate (5,000 cells/well) and the MTT solution (Sigma-Aldrich) was used as described elsewhere (36). Spectrophotometric absorbance at 570 nm wavelength was determined by GloMax-Multi Detection System (Promega). Colony formation assay was performed by using a 6-well plate pre-coated with 1% of soft agar SeaKEM LE Agarose (LONZA) dissolved in medium, supplemented by 1X glutamine, 1X antibiotics, 20% FBS and 800 ng/mL of neomycin. The 3000 cells/mL were plated on the upper layer (0.7% agarose dissolved in medium plus 1X glutamine, 1X antibiotic, 20% of FBS and 800 ng/mL of neomycin). This top layer was covered by 1 mL of complete medium. The cell colonies were fixed with 10% Methanol/10% Acetic Acid for 10′ and then stained with a 0.005% Crystal Violet (Sigma-Aldrich).

Cell migration was analyzed by wound-healing assay. Briefly, an opportune amount of cells were grown in 6-well plates. Wound injury was made with the tip of a sterile micropipette and cells were allowed to migrate for up to 48 hours. In vitro invasion assay was performed using a 24-well Transwell insert (8-μm-pore size) pre-coated with BD Matrigel matrix (BD Biosciences; ref. 37). The invading cells were fixed with PFA 4%, rinsed with PBS, permeabilized with EtOH 100%, stained with 1% Crystal Violet and photographed. Cells were quantified as the average number of cells found in five random microscopic fields in three independent inserts.

To establish xenograft tumors, 1 × 10⁷ HCT15 cells, stably transfected with V5-Jag1-ICD expressing vector or negative control, were, respectively, injected subcutaneously into right and left dorsal flank of CD1 nude mice (n = 6). Conversely, 2 × 10⁶ DLD1 cells were injected subcutaneously into the hind leg of 6-week-old CD1 nude female (n = 6). When tumor reached a mean volume of 150 mm³, the animals were randomly separated into different groups and treated, respectively, with 5FU at 40–50 mg/kg/2–3 days intraperitoneally (n = 4), U0126 25 μmol/kg/2 days intraperitoneally (n = 4) and TAPI-2 at 2 mg/Kg/2 days oral gavage (n = 6), dissolved in 0.2 mL of saline solution. The control group received injection/oral gavage of vehicle alone. After 27 days, mice were killed and tumors were excised. Tumor size was measured every 3/4 days with a caliper and volume was calculated according to the formula: length × width × 0.5 ×(length+width; ref. 38). Harvested tumor tissues were subjected to RNA and WCE extraction as described.

Samples from the following cohorts: 545 patients with colorectal cancer (GEO ID: gse3958283; ref. 39) and 83 patients with colorectal cancer (GEO ID: gse28702; ref. 40) were selected and analyzed for the Jagged1 gene expression levels. The expression values of Jagged1 were filtered in each analysis using the expression probe set 209099_x_at. The expression value of Jagged1 is given in log₂ scale after normalizing data with rma and mas5.0 normalization. GraphPad Prism 6 was used for statistical analysis and P values were calculated using Student t test and one-way ANOVA, where appropriate.

All results were confirmed in at least three independent experiments and all quantitative data were reported as the mean ± SD. Student t or ANOVA tests for unpaired samples were used to assess differences among groups. A P value of <0.05 was considered statistically significant (n.s., nonsignificant, P > 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001, and ****, P < 0.0001).

On the basis of the observation that Jagged1 transcripts are overexpressed in a large number of human colorectal cancers, while they are undetectable in the adjacent normal tissue (41, 42), we monitored the expression of Jagged1 transcripts in several human colorectal cancer cell lines by qRT-PCR assays. Accordingly, we found a significant upregulation of Jagged1 mRNA in most colorectal cancer cell lines, compared with the normal colon cell line CCD-18Co (Fig. 1A), being HT29 and RKO cells the only exceptions. It is well demonstrated that the transmembrane Jag1-FL undergoes ADAM17-mediated ectodomain processing, resulting in the Jag1-ECD shedding, followed by PS/γ-secretase–dependent intramembrane proteolysis that releases an intracellular fragment (Jag1-ICD; refs. 19–21, 23). Intriguingly, here we provide the first evidence of a Jag1-FL aberrant processing in colorectal cancer cell lines, which ultimately results in the release of a remarkable amount of Jag1-ICD (Fig. 1B), able to translocate into the nucleus, as revealed by subcellular protein fractionation (Fig. 1C). Notably, as suggested by Supplementary Fig. S1, Jagged1 is strongly expressed/processed only in colorectal cancer cell lines presenting simultaneously APC-β-catenin/Kras.

Because Jag1-ICD might have a role in tumor development and carcinogenesis (20, 23), we explored whether its overexpression might affect oncogenic properties of colorectal cancer cells. We found that Jag1-ICD ectopic expression in HCT15 cells (Supplementary Fig. S2A), which express low levels of endogenous Jagged1 (HCT15-V₅Jag1-ICD), determined a significant increase in cellular proliferation, as revealed by the MTT assay (Fig. 2A), induced an increased clonogenic capacity in soft agar colony formation assays (Fig. 2B) and sustained cell invasion activity in vitro, using Transwell inserts (Fig. 2C). Interestingly, Jag1-ICD overexpression was also able to sustain colorectal cancer cells invasion/migration ability, as demonstrated by wound-healing assays (Fig. 2D). This was associated with an increased expression of invasion-related snail and mmp9 genes, as revealed by qRT-PCR (Fig. 2E).

To further validate these in vitro results, we xenografted HCT15-V₅Jag1-ICD- or pcDNA₃-Neo empty vector-transfected HCT15 cells, into nude mice. Twenty-seven days after injection, we found that Jag1-ICD expressing clones generated larger tumors when compared with control cells (Fig. 2F and G). Importantly, this was associated to an increased expression of mmp9, snail1, snail2, cyclin D2, and PCNA transcripts in Jag1-ICD tumors, when compared with controls (Fig. 2H). In addition, Fig. 2I shows a strong positivity for Jagged1 immunostaining in human primary colon cancer specimens and this is consistent with our preclinical data.

Overall, these data indicate that the constitutive expression of Jag1-ICD enhances the tumorigenic behavior of colorectal cancer cells, suggesting that Jag1-ICD possesses an intrinsic oncogenic activity.

So far, our results support an intrinsic oncogenic activity of Jag1-ICD, possibly impinging on an invasion/migration phenotype, which is typically associated with epithelial–mesenchymal transition (EMT). This is consistent with in silico analysis of a public dataset (40), which reveals increased Jagged1 expression in patients with metastatic colorectal cancer compared with primary tumors (Fig. 3A). Because PMA is known to support EMT with effects on cell migration and tumor formation in colorectal cancer cells (43), we assessed the potential role of Jag1-ICD in this context. Noteworthy, PMA-treated colorectal cancer cell lines readily acquired a spindle-shaped morphology consistent with mesenchymal transition (Fig. 3B), associated with a strong upregulation of snail1 and snail2, vimentin and N-cadherin and a downmodulation of E-cadherin observed at the mRNA and/or protein levels (Fig. 3C and D). Interestingly, immunoblotting also revealed a time-dependent increase of cleaved Jag1-ICD in PMA-treated HCT15, SW948 and DLD1 cells (Fig. 3D). Altogether these observations support a correlation between Jag1-ICD accumulation and PMA-induced EMT in colorectal cancer cell lines. Indeed, siRNA-mediated Jag1 depletion (Supplementary Fig. S2B) significantly compromised the migratory activity of Jag1-silenced HCT15 cells both under basal (DMSO; decreased by 40%) or PMA-induced conditions (decreased by 30%; Fig. 3E) and significantly impaired snail mRNA expression (Supplementary Fig. S2C). We previously demonstrated that Jag1-ICD directly interacts with CSL/RBP-Jκ transcription factor, sustaining its transcriptional activation (23). Sequence analysis of the human snail1 and snail2 promoters identified consensus CSL/RBP-Jκ–binding sites (Supplementary Fig. S2D). ChIP assays around these sites showed a significant recruitment of CSL/RBP-Jκ and Jag1-ICD in PMA-treated HCT15 cells (Fig. 3F).

Overall, these findings demonstrate that nuclear accumulated Jag1-ICD directly controls the expression of EMT-related genes and the migratory activity of colorectal cancer cells, unveiling a tight link between aberrant Jagged1 processing and colorectal cancer aggressiveness.

Jagged1 is a substrate of the catalytic activity of ADAM17 that allows the shedding of Jag1-ECD ectodomain, an obligatory step before the cleavage of Jag1-ICD by the PS/γ–secretase complex (19). It is known that PMA enhances ADAM17 sheddase activity, by directly inducing Erk kinase phosphorylation and activation (44, 45), which is an important prerequisite for ADAM17 triggering (44, 46).

First, to assess the phosphorylation status of ADAM17 upon PMA treatment in colorectal cancer cells, we carried out immunoprecipitation assays of endogenous proteins from DLD1 cell line. As shown in Fig. 4A, we revealed a rapid induction of Ser-phosphorylation on ADAM17 within 15 minutes of stimulation. Then, we investigated the effects of PMA on ADAM17 sheddase activity, by monitoring Jagged1 cleavage in colorectal cancer cell lines. Interestingly, PMA treatment induced extensive Jagged1 processing, revealed by a significant increase of soluble Jag1-ECD and Jag1-ICD fragments in HCT15, LoVo, SW948 and DLD1 colorectal cancer cell lines with different expression levels of Jagged1, associated to an important Erk activation (Fig. 4B). Consistently, Erk inhibition via the U0126 antagonist strongly impaired Jagged1 processing, indicating that Jag1-ICD accumulation is Erk-dependent (Fig. 4C). Notably, ectopic Jag1-ICD, stably transfected in HCT15-V₅Jag1-ICD cells, is sufficient to revert the effect of U0126, as revealed by the sustained activation of EMT-linked target genes, as snail and E-cadherin (Supplementary Fig. S2E and S2F). In silico analysis of a public dataset (39), considering a large cohort of patients with colorectal cancer, showed that increased expression of Jagged1 transcripts is significantly associated to Kras mutation-bearing samples compared with Kras wt tumors (Fig. 4D). Moreover, it is reported that oncogenic Kras is able to regulate ADAM17 activity in a MEK/ERK-dependent manner (7). Interestingly, Jagged1 is strongly processed only in colorectal cancer cell lines bearing Kras mutations (Fig. 1B; Supplementary Fig. S1). These observations support the existence of a direct correlation between the aberrant activation of the Kras/Erk pathway and the Jagged1 processing in colorectal cancer cells. To clarify this correlation, we investigated the status of Jagged1 protein in response to siRNA-mediated Kras depletion in HCT15, SW948 and DLD1 colorectal cancer cell lines. Kras silencing resulted in a marked impairment of Jagged1 processing, revealed by a significant decrease in Jag1-ECD shedding and Jag1-ICD release, strongly suggesting a direct link between Kras activity and Jagged1 processing (Fig. 4E). Consistently, the overexpression of mutant Kras (pBabe Kras 12V), by retroviral infection of CCD18-Co cells (CCD18-Co-Kras cell line), causes a drastic change in cell morphology with the appearance of spindle-shaped cells, compared with empty backbone infected cells (pBabe-Puro; Fig. 4F). Notably, pERK was strongly induced by Kras in CCD18-Co-Kras cell line, which triggers Jag1-ICD release, compared with CCD18-Co-Puro cells (Fig. 4G, left). Of note, the Kras-induced Jag1-ICD processing was inhibited by TAPI2 compound (Fig. 4G, right).

Altogether these results highlight a Kras/Erk/ADAM17/Jagged1 signaling axis in colorectal cancer cells, whereby Kras activation leads to Erk-ADAM17-dependent Jagged1 cleavage, resulting in the nuclear accumulation of Jag1-ICD.

To explore the role of Jag1-ICD in sustaining the tumorigenic potential of colorectal cancer cells, we abrogated constitutive Jagged1 cleavage in HCT15 cells by using the TAPI-2 compound, which is able to inhibit ADAM17 activity (Fig. 5A, left). TAPI-2 treatment impaired HCT15 cell growth by 40%, as determined by trypan blue cell counting (Fig. 5A, right), associated to a G₀–G₁ cell-cycle arrest (Fig. 5B). This was associated to the decrease of the endogenous cyclin D2 and PCNA transcripts, as revealed by qRT-PCR, in TAPI-2–treated when compared with control cells (Fig. 5C). Consistent with previous results, the inhibition of Jag1-ICD release by TAPI-2 significantly decreased HCT15 invasiveness through the Matrigel (56%; Fig. 5D) and reduced the expression of invasion-related transcripts such as mmp9, snail1, and snail2 (Fig. 5E). In addition, we carried out wound-healing assays to determine the biological effect of Jag1-ICD on the migration capability of HCT15 cells, treated with PMA alone or co-treated with TAPI-2 compound (Fig. 5F). Notably, treatment with PMA alone strongly determined a time-related increased motility (increased by 40%), when compared with control. Intriguingly, PMA effect is delayed in presence of TAPI-2 compound (decreased by 30%), which counteracts the PMA-induced Jag1-ICD shedding, indicating that Jag1-ICD release is required for colorectal cancer cell migration. Of note, TAPI-2 treatment does not have any impact on HCT15-V₅Jag1-ICD cells, expressing Jag1-ICD constitutively, revealed by sustained expression levels of EMT-target genes (Supplementary Fig. S2G and S2H). To investigate the effects of Jag1-ICD on colon cancer in vivo, DLD1 cells were injected subcutaneously into the flanks of nude mice, which were treated with TAPI-2 or control vehicle. The results in Fig. 5G and H show that tumor volume was clearly decreased in TAPI-2 treated with respect to control mice. Western blot results from tumor xenografted samples showed that the expression levels of Jag1-ICD were markedly decreased in samples obtained from mice treated with TAPI-2 (Fig. 5I).

Altogether these results indicate that Jag1-ICD plays a role in regulating malignant features, such as proliferation and invasion/migration ability in colorectal cancer cell lines both in vitro and in vivo.

Kras mutation is an important predictor of drug resistance in several cancers and is associated with a worse prognosis (8). Notably, it is reported that chemotherapy results in a significant increase of ADAM17 activity and growth factors shedding, which determine drug resistance in Kras^mut colorectal cancer tumors (4, 47). Chemoresistance is often associated to acquisition of EMT (48), the phenotype induced in colorectal cancer cells by the Kras/Erk/ADAM17/Jag1-ICD axis that we described above. Interestingly, it is reported that high Jagged1 expression levels, combined with low E-cadherin expression, in cancer cells of patients with colorectal cancer are correlated with poor prognosis, poorer survival rate and increased risk of recurrence (13). Overall these observations allowed us to speculate about a direct link between an enforced Jag1-ICD shedding and the acquisition of resistance. In keeping with this hypothesis, stable Jag1-ICD overexpression in HCT15 cells was sufficient to confer resistance to 5FU and irinotecan agents, as revealed by a sustained survival rate in colorectal cancer cells, with respect to untransfected cells (Fig. 6A). To explore the possibility that the resistance to 5FU and/or irinotecan may depend on Jag1-ICD, we tested the impact of both chemotherapic agents on Jagged1 processing. Surprisingly, treatment of HCT15 cells with 5FU (Fig. 6B) or irinotecan (Fig. 6C) for 24 hours increased the release of Jag1-ICD in a dose-dependent manner, associated to an increased phosphorylation status of Erk and ADAM17 (Fig. 6B and C) and correlated with the modulation of the EMT-specific markers mmp9, snail1, snail2, and E-cadherin (Fig. 6D). Notably, 5FU- or irinotecan-induced Jag1-ICD processing was significantly decreased by TAPI-2 (Fig. 6E) or by U0126 (Fig. 6F) compounds. Interestingly, Supplementary Fig. S3A–S3E, shows that the effects above described are also observed in DLD1, a cell line with high expression levels of endogenous Jag1-ICD. To confirm such in vitro results, we xenografted DLD-1 cells into nude mice, treated with 5FU or U0126 alone or in combination (5FU/U0126) and the tumor growth was measured along the time (Fig. 6G). A significant reduction of Jag1-ICD levels was observed in tumors treated with U0126 alone, associated to a drastic reduction of tumor growth (Fig. 6H and I), with respect to vehicle-treated mice. As expected, no significant difference was found in the tumor size from mice treated with 5FU alone, which further increases the release of endogenous Jag1-ICD, when compared with control mice (Fig. 6G–I), sustaining the idea that 5FU is able to induce colorectal cancer resistance by inducing Jag1-ICD shedding. On the basis of the compelling in vitro evidences, U0126 is not able to completely counteract the 5FU-dependent Jag1-ICD increase in tumors from mice with a combined treatment, 5FU/U0126 (Fig. 6G–I).

Overall these data demonstrate that 5FU and irinotecan are able to strongly sustain the Jagged1 processing, by triggering the Erk/ADAM17 axis, which results in the release of the Jag1-ICD oncogenic fragment, able to confer chemoresistance to colorectal cancer, both in vitro and in vivo.

The Notch ligand Jagged1 is upregulated in a large number of cancers, where it plays a key role in cell growth, EMT and metastatic process (10). An increased expression of Jagged1 has been identified in about 50% of human colorectal cancer (41) where it has been correlated with poor prognosis and recurrence (13). To date, the most widely accepted scenario suggest that the increased expression of Jagged1 ligand identified in colorectal cancer triggers an overactivation of Notch signaling (42, 49). However, Jagged1 may be processed in a fashion similar to Notch receptor, ultimately resulting in the release of the nuclear-targeted intracellular domain Jag1-ICD, thus triggering a reverse signaling (19, 23). Herein, we demonstrate that Jag1-ICD is able to empower the Kras-mediated oncogenic signaling, by sustaining features of malignancies, tumor-cell invasion, migration, and resistance to chemotherapy.

Previous data revealed that more than one oncogenic “driver” is deregulated in colorectal cancer tumors (1). Mutations in the Wnt pathway cause colon cancer through constitutive activation of the β-catenin/TCF transcription complex (50). Recent reports have shown that β-catenin/TCF is responsible of a direct regulation of Jagged1 expression, which is required for tumorigenesis in the intestine (9). In addition, gain-of-function mutations in RAS gene are present in approximately 50% of colon cancers (1, 5). Notably, oncogenic Kras signaling increases the β-catenin stability, by modulating its phosphorylation at serine 552 (51). Interestingly, increasing evidence suggests that the oncogenic Kras mutations control ADAM17 activity and growth factor shedding, via regulation of MEK/Erk/Adam17 signaling axis (4). These results are supported by the observation that Erk activation phosphorylates and associates with ADAM17 (7, 46). In agreement with these data, we provide the first evidence that Kras^mut colorectal cancer cells specifically show increased expression of Jagged1, which is constitutively processed by ADAM17, in a Kras-dependent manner. Of note, we show on one side that Kras-silencing attenuates significantly the Jag1-ICD release and on the other side that Kras ectopic expression directly empowers the Jagged1 cleavage, supporting the idea that Jagged1 processing is a novel substrate of Kras signaling in colorectal cancer cells. Here, we demonstrate that the constitutive Jagged1 cleavage observed in colorectal cancer cells is dependent upon Erk activation, able to phosphorylate ADAM17, as revealed by PMA stimulation or on inhibition of Erk activity with U0126 compound, both in vitro and in vivo experiments. Noteworthy, the aberrant PMA-induced Jag1-ICD release is associated to a marked increase of EMT markers Snail, vimentin, N-cadherin, and E-cadherin. On the other side, TAPI-2–mediated ADAM17 inhibition correlates with different biological outcomes, including significant decrease of cell growth and reduction of migration and invasion phenomena, both in vitro and ex vivo, in tumor xenografts experiments. Previous observations suggest that Jag1-ICD may directly interact with RBP-Jκ transcription factor (23). For the first time, we demonstrate that Jag1-ICD is able to trigger an intrinsic reverse signaling by regulating snail1 and snail2 promoter activity, via CSL/RBP-Jκ. Moreover, pre-clinical studies, performed by using HCT15-V₅Jag1-ICD xenografts experiments, sustain the idea that the persistent expression of Jag1-ICD plays an oncogenic function also in in vivo models.

Interestingly, Kras mutations are often associated with a colorectal cancer worse prognosis (4, 7, 8). Of note, Kras status has been correlated with Jagged1 expression in patients with colorectal cancer and associated with a poorer survival rate and increased risk of recurrence, characterized by low cadherin expression and the induction of EMT, but the molecular mechanism is unknown (13). In addition, it has been reported that current chemotherapy acutely activates ADAM17 that plays an important role in drug resistance in colorectal cancer tumors (7, 47). Emerging evidence associates chemoresistance with the development of an EMT-like phenotype in cancer cells (52), suggesting that EMT, metastasis, and chemoresistance are closely related to each other in tumor progression (48). In accordance with these observations, we show that the 5FU or irinotecan treatments increase the endogenous Jag1-ICD release, via Erk phosphorylation, in vitro or in xenografts experiments and are able to induce EMT, as revealed by modulation of endogenous specific markers. Therefore, our data indicate that the constitutive processing of Jagged1, induced by 5FU or by irinotecan, could be a crucial event correlated with increased risk of recurrence, poor outcome and resistance to chemotherapy of Kras^mut colorectal cancer.

In conclusion, we provide evidence that Jagged1 is not only abundantly expressed but is also constitutively processed in colorectal cancer Kras molecular subtype tumors, via a Kras/Erk/ADAM17 pathway. The release of Jag1-ICD, in turn is able to empower the oncogenic Kras signaling pathway, via a novel mechanism, which sustains invasion and contributes to chemoresistance. Therapies targeted at this definite pathway may provide a novel method to sensitize and/or to disrupt the resistance mechanism of Kras-mutated colorectal cancer to chemotherapy, to finally improve overall tumor control and reduce tumor recurrence.

No potential conflicts of interest were disclosed.

Conception and design: M. Pelullo, D. Bellavia

Development of methodology: M. Pelullo, F. Nardozza, S. Zema, R. Quaranta, R. Palermo, C. Capalbo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Nicoletti, M.P. Felli, C. Capalbo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Pelullo, Z.M. Besharat, B. Cerbelli, G. d'Amati, C. Capalbo, C. Talora, L. Di Marcotullio, S. Checquolo, I. Screpanti, D. Bellavia

Writing, review, and/or revision of the manuscript: Z.M. Besharat, B. Cerbelli, G. d'Amati, C. Capalbo, G. Giannini, S. Checquolo, I. Screpanti, D. Bellavia

Study supervision: I. Screpanti, D. Bellavia

This article is dedicated to the memory of Prof. Alberto Gulino. This work has been supported by Italian Ministry of Education, University and Research—Dipartimenti di Eccellenza—L. 232/2016, Associazione Italiana Ricerca Cancro (AIRC) Grants # IG20801 (to L. Di Marcotullio), #IG17734 (to G. Giannini); by Sapienza University Project Num: RP1181643121DD86 (to D. Bellavia); RG116154E2C7A6FB and MIUR PNR 2015-2020 ARS01_00432 (to I. Screpanti).

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.