Abstract

   Wnt/β‐catenin/transcription factor (TCF) transcriptional activity plays
   an integral role in colorectal cancer (CRC) carcinogenesis. However, to
   date, no drugs targeting this pathway are used in clinical practice
   owing to the undesirable and serious side effects. In this study, it is
   found that the transcriptional regulation of Wnt pathway is activated
   and associated with liver metastasis in CRC. Through high‐throughput
   screening of 24 inhibitors on 12 CRC and three colorectal organoids in
   this organoid living biobank, adavivint is found to exhibit anti‐tumor
   activity and low toxicity in colorectal organoids, independent of the
   canonical Wnt/β‐catenin signaling. Mechanistically, ADAM10 is screened
   as a target of adavivint to specifically regulate the protein
   expression of NOTCH2, which mediates the transcriptional regulation of
   the Wnt pathway. NOTCH2 not directly interact with TCF7‐like 2
   (TCF7L2), a key downstream transcriptional factor of canonical
   Wnt/β‐catenin signaling, but directly activated the transcription of
   TCF7L2 and Wnt target genes, such as MYC, JUN and CCND1/2. Furthermore,
   use of adavivint or blockage of ADAM10/NOTCH2/TCF7L2 signaling enhances
   the chemosensitivity of CRC cells. Overall, this study provides a
   promising candidate for the development of small‐molecule inhibitors
   and reveals a potential therapeutic target for CRC.

   Keywords: ADAM10, colorectal cancer, liver metastasis, NOTCH2,
   organoid, TCF7L2, Wnt pathway
     __________________________________________________________________

   Transcriptional regulation of Wnt pathway is activated and associated
   with liver metastasis in CRC. By high‐throughput screening of
   inhibitors with CRC organoids, adavivint is found to suppress tumor
   growth and enhance chemotherapy sensitivity of CRC by targeting
   transcriptional regulation of Wnt pathway. Further exploring the action
   mechanisms of adavivint identifies ADAM10/NOTCH2/TCF7L2 signaling as a
   potential therapeutic target for CRC.

   graphic file with name ADVS-12-2405758-g005.jpg

1. Introduction

   Colorectal cancer (CRC) is among the most prevalent malignancies and a
   primary contributor to cancer‐related mortality.^[ [50]^1 ^]
   Approximately 50% of patients diagnosed with CRC develop liver
   metastases^[ [51]^2 ^] with a 5‐year survival rate of less than 10%.^[
   [52]^3 ^] Surgical intervention, systemic chemotherapy, and targeted
   therapies, such as those involving cetuximab and bevacizumab, are the
   primary treatment options for CRC liver metastases. Owing to its
   considerable heterogeneity and complexity, CRC exhibits varied
   responses to chemotherapy and targeted therapies, and acquired drug
   resistance during treatment significantly impedes therapeutic efficacy.
   Improving the prognosis of patients with chemotherapy‐resistant
   metastatic CRC remains a challenge. Therefore, novel therapeutic
   targets are urgently needed to overcome drug resistance in CRC
   treatment.

   CRC is caused by the accumulation of multiple gene mutations, such as
   APC, TP53, and KRAS.^[ [53]^4 , [54]^5 ^] Specifically, APC
   mutation‐mediated overactivation of the Wnt signaling pathway is a key
   driver of CRC carcinogenesis, with approximately 80% of CRC cases
   harboring APC mutations.^[ [55]^4 , [56]^6 ^] Mechanistically, APC
   mutations accelerate the nuclear accumulation of β‐catenin, leading to
   its binding with transcription factor (TCF)/lymphoid enhancer factor
   (LEF) and activation of transcription for downstream target genes of
   the classical Wnt pathway, including MYC, JUN, etc.^[ [57]^6 , [58]^7 ,
   [59]^8 , [60]^9 ^] Additionally, activation of Wnt pathway also
   promotes CRC progression by influencing the tumor microenvironment,
   cell proliferation, apoptosis, autophagy, metabolism, invasion, and
   migration.^[ [61]^10 , [62]^11 ^] Furthermore, Wnt pathway can
   crosstalk with the Notch and Sonic Hedgehog pathways, presenting
   significant implications for cancer therapy.^[ [63]^12 , [64]^13 ^]
   Although the Wnt pathway plays a crucial role in tumorigenesis and
   tumor development, an increasing number of studies have reported that
   it is regulated by other pathways, including the KRAS, phosphoinositide
   3‐kinase–Akt, and Notch pathways.^[ [65]^14 ^] These findings indicate
   that targeting crucial upstream regulatory components may be a
   desirable approach for treating Wnt pathway‐active CRC.

   Despite the critical role of the Wnt signaling pathway in
   carcinogenesis and tumor development, there is currently no approved
   drug targeting this pathway.^[ [66]^15 , [67]^16 ^] Several inhibitors
   targeting the canonical Wnt/β‐catenin pathway, including porcupine
   O‐acyltransferase (PORCN) inhibitors and anti‐R‐spondin (RSPO)
   antibodies, are currently undergoing clinical trials.^[ [68]^15 ^] The
   Wnt signaling pathway also mediates resistance to conventional
   chemotherapy and radiotherapy.^[ [69]^10 ^] Inhibition of Wnt ligand
   secretion by PORCN inhibitors or RSPO3 blockade leads to rapid and
   sustained tumor regression.^[ [70]^17 , [71]^18 , [72]^19 ^] The
   canonical Wnt/β‐catenin pathway regulates cell proliferation,
   development, self‐renewal, and differentiation. However, Wnt pathway
   inhibitors also exerts on‐target adverse effects.^[ [73]^16 ^] For
   instance, blocking the Wnt pathway using tankyrase inhibitors induces
   severe gut toxicity in mice used for cancer treatment, thereby limiting
   the application of tankyrase inhibitors for human patients ^[ [74]^17 ,
   [75]^19 ^]. Although PORCN inhibitors do not exhibit significant side
   effects in the guts of tumor‐burdened mice, they induce various
   on‐target dose‐limiting toxicities, including bone fractures.^[ [76]^20
   ^] Using a 3D‐organoid model, some researchers have demonstrated that
   70% of CRC cases bearing APC mutations exhibit a favorable growth
   status independent of the Wnt pathway activators,^[ [77]^20 , [78]^21
   ^] which further limits the efficacy of Wnt pathway inhibitors. Despite
   the potential of the pharmacological inhibition of Wnt signaling as a
   therapeutic option for patients with cancer, finding effective and
   low‐toxicity inhibitors of this pathway remains a significant
   challenge. Notably, some inhibitors targeting the Wnt pathway, such as
   adavivint and teplinovivint, have advanced to phase II clinical
   trials.^[ [79]^22 ^] These compounds suppress the Wnt signaling pathway
   through other bypasses or upstream regulatory pathways, thereby
   exerting fewer side effects.

   Recently, 3D‐organoid culture technology for tumors has been developed
   that closely recapitulates the properties of the original tumor and
   provides significant advantages for biological and drug research on
   cancer.^[ [80]^23 ^] Han et al. revealed that CRC organoids bearing
   mutations in the transforming growth factor‐beta, TP53, and Hippo
   pathways exhibit resistance to Wnt pathway inhibitors,^[ [81]^20 ^]
   suggesting the need for pathway‐targeted treatment. In this study,
   through transcriptomic analysis of CRC organoids and tissues, we found
   that the transcriptional activation of Wnt pathway in CRC was
   independent of the canonical Wnt/β‐catenin pathway and associated with
   liver metastasis. Subsequently, high‐throughput screening of inhibitors
   targeting the Wnt pathway using an organoid model revealed that
   adavivint not only suppressed CRC organoid growth but also showed low
   toxicity to colorectal organoids. Mechanistically, adavivint blocked
   the transcription of Wnt target genes by bypassing ADAM10/NOTCH2
   signaling, resulting in growth arrest and susceptibility to
   chemotherapy. This study provides a promising small‐molecule inhibitor
   and therapeutic target for CRC.

2. Results

2.1. Transcriptional Regulation of the Wnt Pathway, Independent of the
Canonical Wnt/β‐Catenin Pathway, is Activated and Associated with Liver
Metastasis in CRC

   Organoid culture could obtain high‐purity cancer cells compared with
   organoid‐derived tumor tissue and favor studies focusing on cancer
   cells by excluding disruption of the microenvironment. We successfully
   generated six primary CRC organoids from six primary CRC samples
   (Figure [82] 1A), nine liver metastatic CRC organoids from eight liver
   metastatic CRC samples (Figure [83]1B), and four colorectal organoids
   from four normal colorectal tissues (Figure [84]1C). Hematoxylin and
   eosin (HE) staining of the organoids and corresponding tissues was
   performed for pathological identification. Transcriptomic (Figure
   [85]S1A, Supporting Information) and whole exome sequencing analyses
   (Figure [86]S1B, Supporting Information) were performed to demonstrate
   that the organoids preserved the transcriptomic and genomic
   characteristics of the derived tumor tissues. Over 90% of the CRC
   organoids showed either TP53 or APC mutations (Figure [87]1D). Through
   gene set variation analysis (GSVA) using RNA‐seq data, enrichment score
   of Wnt pathway (name as “WNT score”) could be calculated. By performing
   a correlation analysis of WNT scores between CRC tissues and CRC
   organoids, we found that CRC organoids could mimic the transcriptional
   activity of the Wnt pathway in CRC tissues (Figure [88]1E). We observed
   higher transcriptional activity of the Wnt pathway in CRC organoids
   than in colorectal organoids; however, transcriptional regulation of
   the Wnt pathway was not specifically activated in APC‐mutant CRC
   organoids (Figure [89]1F). On the Cancer Genome Atlas (TCGA) database,
   enrichment scores of the Wnt pathway in APC wild‐type and APC‐mutant
   CRC tissues were similar (Figure [90]1G). Further analysis revealed a
   higher transcriptional activity of the Wnt pathway in liver metastatic
   CRC than in colorectum or primary CRC (Figure [91]1H). The protein
   levels of Wnt pathway members in five primary CRC organoids and nine
   liver metastatic CRC organoids were detected by western blotting.
   Further quantitative analysis revealed elevated levels of WNT3A, LEF1,
   and JUN in liver metastatic CRC organoids (Figure [92]1I). We noted
   that the P2 and LM8 organoids with high WNT scores expressed little
   β‐catenin (Figure [93]1I). We then used the medium without the Wnt
   pathway activator RSPO1 to culture CRC organoids and found that these
   organoids grew well in the medium minus RSPO1 (Figure [94]1J).
   Interestingly, the P2 and LM8 organoids grew faster in the medium
   without RSPO1 than in the medium containing RSPO1. RNA‐seq and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) pathway analyses revealed that
   the P2 and LM8 organoids showed enhanced transcriptional regulation of
   the Wnt and other tumor‐associated pathways, especially the Notch and
   Jak/Stat pathways, in the medium without RSPO1 (Figure [95]1K). These
   results suggest that transcriptional regulation of Wnt pathway is
   activated in CRC, especially in liver‐metastatic CRC, and regulated
   independent of the canonical Wnt/β‐catenin pathway.

Figure 1.

   Figure 1
   [96]Open in a new tab

   Transcriptional regulation of Wnt pathway is activated and associated
   with liver metastasis in CRCs. A–C) Establishment of primary
   CRC‐derived, liver‐metastatic CRC‐derived and colorectal organoids.
   Representative bright‐field microscopy images of six organoids derived
   from primary CRC tissues (P1‐P6), nine organoids derived from
   liver‐metastatic CRC tissues (LM1‐LM8) and four colorectal organoids
   derived from colorectal tissues (C1N,C3N,C6N,C7N), and HE staining of
   these organoids and corresponding tissues. D) Top 10 mutated genes in
   established organoids by WES sequencing. E) Correlation analysis
   reveals positive relationships of Wnt score between organoids and
   paired tumor tissues (n = 14). F) Expression of enrichment scores of
   Wnt pathway in colorectal organoids (n = 4), APC wild‐type CRC
   organoids (n = 4) and APC‐mutant CRC organoids (n = 8). G) Expression
   of enrichment score of Wnt pathway in APC wild‐type (n = 142) and
   APC‐mutant CRC tissue (n = 391) in TCGA database. H) Expression of
   enrichment scores of Wnt pathway in liver‐metastatic CRC tissues (n =
   9), liver tissues (n = 4), normal colorectal tissues (n = 7) and
   primary CRC tissues (n = 7), or in colorectal organoids (n = 4),
   primary CRC organoids (n = 6) and liver‐metastatic CRC organoids (n =
   9). I) Western blot analysis of members of classical Wnt pathway in 5
   primary CRC organoids and 9 liver‐metastatic CRC organoids (Left). WNT
   scores and APC mutation are also displayed. Symbol “U” indicates the
   mutation is unknown. Heatmap indicates the relative protein levels of
   β‐catenin, TCF7L2, JUN, WNT3A and LEF1 (Right). J) Bright field
   photograph of the indicated CRC organoids cultured in medium with or
   without R‐spondin1. K) Heatmap shows changed enrichment scores of the
   indicated pathways after treated by medium minus R‐spondin1. Changed
   enrichment score = enrichment scores of pathways in organoids cultured
   medium minus R‐spondin1 – enrichment scores of pathways in organoids
   cultured medium plus R‐spondin1. “MT” or “WT” represents “mutation” or
   “wild type” respectively. WNT scores indicate Wnt pathway enrichment
   scores by GSVA analysis. Bar: 40×, 200µm; 200×, 50µm. All P‐values in
   (F, G, and H) are calculated using an unpaired two‐sided Student's
   t‐test, and data presented as mean ± SD. *p < 0.05, **p < 0.01, ***p <
   0.001.

2.2. High‐Throughput Screening Using CRC and Colorectal Organoid Models
Revealed Adavivint as a Therapeutic Small‐Molecule Inhibitor Targeting the
Wnt Pathway

   Given the proven specificity of the activated transcription of the Wnt
   pathway in CRC, we searched for 24 inhibitors targeting this pathway
   (Table [97]S1, Supporting Information). Among these compounds, we found
   that some suppressed the Wnt pathway were nor clearly clarified, such
   as adavivint and teplinovivint. These compounds may suppress the Wnt
   pathway at the transcriptional level through other pathways.
   Figure [98] 2A provides detailed information on our screening process
   using established organoids, and the heatmap in Figure [99]2B
   illustrates the relative viability of the 24 inhibitors after a 96‐h
   treatment at a concentration of 1 µM. Of the 24 small‐molecule
   inhibitors, tegatrabetan, adavivint, teplinovivint, CWP232228 and DK419
   exhibited the highest inhibitory activity against CRC organoids. JW74
   and MSAB showed slight inhibitory effects on CRC organoids, and other
   19 inhibitors did not showed toxicity on CRC organoids, including
   pamidronic acid, and KY1220. To explore the action mechanisms of these
   compounds, we conducted RNA‐seq analysis on CRC organoids treated with
   tegatrabetan, adavivint, teplinovivint, CWP232228, DK419 at 1 µM for 24
   h. Two inhibitors, pamidronic acid, and KY1220 who did not inhibit the
   growth of CRC organoids, were also included. GSVA of the KEGG pathway
   revealed that tegatrabetan, adavivint and teplinovivint suppressed
   transcriptional regulation of the Wnt pathway. Among the pathways known
   to be involved in tumor development, tegatrabetan and adavivint
   effectively inhibited the Hedgehog and Notch pathways (Figure [100]2C).
   However, the administration of CWP232228 and DK419 reversed the changes
   in the transcription levels of these pathways. For example, CWP232228,
   specifically antagonizing binding of β‐catenin to T‐cell factor (TCF)
   in the nucleus, activated transcriptional activation of Wnt pathway and
   Notch pathway. Subsequently, we conducted a more detailed examination
   of the inhibition ratios of various concentrations of tegatrabetan,
   adavivint, teplinovivint, CWP232228, and DK419 in colorectal, primary
   CRC, and metastatic CRC organoids (Figure [101]2D). Area under the
   curve (AUC) values revealed that adavivint not only inhibited CRC
   organoid growth but also showed low toxicity to colorectal organoids
   (Figure [102]2E). Tegatrabetan, specifically disrupting the binding of
   β‐catenin with the scaffold protein transducin β‐like 1, also
   suppressed CRC organoids growth among the five compounds, but caused
   high toxicity on normal colorectal organoids (Figure [103]2E,F).
   RNA‐seq and GSVA revealed that adavivint suppressed the Wnt pathway at
   the transcriptional level in CRC organoids (Figure [104]2G). However,
   nucleus entrance of β‐catenin was not notably attenuated after 24‐h
   treatment with adavivint (Figure [105]2H).

Figure 2.

   Figure 2
   [106]Open in a new tab

   Usage of CRC organoid model to screen inhibitors targeting
   transcriptional regulation of Wnt pathway. A) Process using CRC and
   colorectal organoids to screening inhibitors targeting Wnt pathway. B)
   Heatmap shows relative viability of the 24 indicated inhibitors
   targeting Wnt pathway at the concentration of 1 µM. C) Heatmap shows
   the changed enrichment scores of KEGG pathway in LM3 organoids treated
   by the indicated 7 inhibitors (1 µM) for 24 hours. Changed enrichment
   score = enrichment scores of pathways in inhibitors‐treated organoids –
   enrichment scores of pathways in control organoids. D) Dose‐response
   curves for the 5 indicated compounds in 3 colorectal organoids, 3
   primary CRC organoids and 4 liver‐metastatic CRC organoids. E)
   Histogram shows AUC values of the 5 indicated compounds in CRC
   organoids (n = 7) and colorectal organoids (n = 3). F) Representative
   bright‐field microscopy images of colorectal organoids and CRC
   organoids treated by adavivint and tegatrabetan (1 µM) for 96 hours (n
   = 4). Bar: 40×, 200µm.G) Enrichment scores of Wnt pathway in LM5L, LM5R
   and LM8 organoids treated by adavivint (0.5 µM) for 24 hours (n = 3).
   H) Western blot shows the protein expression of β‐catenin in cytoplasm
   and nucleus after treated by tegatrabetan, adavivint, teplinovivint,
   CWP232228, DK419, pamidronic_acid and KY1220 for 24 hours. All P‐values
   in (E, F and G) are calculated using an unpaired two‐sided Student's
   t‐test, and data presented as mean ± SD. *p < 0.05, **p < 0.01, ***p <
   0.001.

2.3. CRC with High Transcriptional Activity of the Wnt Pathway is Vulnerable
to Adavivint In Vitro and In Vivo

   Next, dose–response curves were plotted for adavivint in CRC organoids
   (Figure [107]S2A, Supporting Information). Correlation analysis showed
   a negative relationship between the enrichment score of the Wnt pathway
   and the AUC of adavivint in tumor tissues and organoids (Figure [108]
   3A). We also established two lung adenocarcinoma (LUAD) and pancreatic
   adenocarcinoma (PDAC) organoids (Figure [109]S2B,C, Supporting
   Information), and the organoids with high transcriptional levels of the
   Wnt pathway were sensitive to adavivint (Figure [110]S2D,E, Supporting
   Information). To explore the inhibitory effect on CRC growth in vivo, a
   patient‐derived organoid xenograft (PDOX) model was constructed using
   CRC organoids with high enrichment scores for the Wnt pathway, and two
   doses of adavivint (5 and 50 mg kg^−1) were administered. Adavivint
   effectively suppressed CRC growth (Figure [111]3B) but did not affect
   the body weight of tumor‐bearing mice (Figure [112]3C) at a dose of 50
   mg kg^−1; similar responses were observed in the two other CRC
   organoids in vivo (Figure [113]3B,C). HE staining and
   immunohistochemistry (IHC) analysis revealed low levels of Ki67 and
   high levels of caspase 3 in the adavivint‐treated PDOX model
   (Figure [114]3D).

Figure 3.

   Figure 3
   [115]Open in a new tab

   Adavivint suppresses CRCs growth in vitro and in vivo. A) Correlation
   analysis between enrichment scores of Wnt pathway and AUC value of
   adavivint in CRC tissues (R = ‐0.73. P = 0.003, n = 14) and organoids
   (R = ‐0.74. P = 0.002, n = 15). Spearman correlation. The effect of
   adavivint on tumor growth B, n = 4) or body weight C, n = 4) of PDOX
   model established using the indicated organoids at the dose of
   5mg kg^−1 or 50mg kg^−1. Data presented as mean ± SD, all P‐values are
   calculated using an unpaired two‐sided Student's t‐test. *p < 0.05. D)
   HE staining and IHC analysis for Ki67 and Caspase3 in representative
   sections from the PDOX model established using the indicated organoids.
   Bar: 200×, 50µm.

2.4. ADAM10/NOTCH2 Signaling is a Target of Adavivint to Regulate the Wnt
Pathway Transcription in CRC

   To comprehensively elucidate the mechanism of adavivint‐mediated growth
   inhibition, we conducted mass spectrometric analysis at 24‐h
   post‐treatment with adavivint. Although 1837 proteins were identified
   (Figure [116] 4A), only six significantly changed proteins were found
   and are displayed in the heatmap (fold change >1.5 or <0.67, P< 0.05;
   Figure [117]4B). CLK2 and DYRK1A, the targets of adavivint reported in
   previous study,^[ [118]^22 , [119]^24 ^] were not detected by mass
   spectrometric analysis in this study. Subsequently, correlation
   analysis was performed to unveil the positive relationship between
   ADAM10 and the Wnt pathway using RNA‐seq data from organoids and
   tissues, along with two additional RNA‐seq datasets from TCGA and GEO
   databases (Figure [120]4C). To further investigate the mechanism by
   which ADAM10 regulates the Wnt pathway, we calculated the enrichment
   scores of 189 KEGG pathways and conducted a correlation analysis
   between these pathways and ADAM10 mRNA levels in CRC organoids (Table
   [121]S2, Supporting Information). Subsequently, two pathways were
   identified, with only the NOTCH pathway showing a significant increase
   in transcriptional levels in liver metastatic organoids and tissues
   (Figure [122]S3A, Supporting Information). Correlation analysis
   revealed positive relationships between the Wnt and Notch pathways in
   CRC tissues and organoids (Figure [123]4D). In the IP assay, we found
   that only ADAM10 directly bound to NOTCH2 but not to other members of
   the NOTCH receptors and ligands (Figure [124]4E). Other ligands of
   NOTCH receptors, such as delta‐like (DLL)‐1, DLL3, DLL4 and Jagged2
   (JAG2) could hardly were not detected by western blotting in CRC
   organoids (data not shown). Among the three members of the ADAM family,
   only ADAM10 was immunoprecipitated using the NOTCH2 antibody
   (Figure [125]4F). Furthermore, western blotting analysis revealed
   significant downregulation of ADAM10 and NOTCH2 levels (Figure [126]4G)
   but no significant changes in NOTCH1, NOTCH4, ADAM9, and ADAM17 levels
   after adavivint treatment for 24 h (Figure [127]S3B, Supporting
   Information). Organoids treated with siRNAs targeting ADAM10 and NOTCH2
   for 72 h showed a significant decrease in the transcriptional levels of
   the Wnt pathway, as determined by RNA‐seq analysis (Figure [128]4H).
   Moreover, downstream genes of the Wnt pathway were significantly
   decreased after adavivint treatment for 24 h, including MYC, JUN, and
   CCND2, whereas the mRNA levels of ADAM10 and NOTCH2 did not exhibit a
   significant decrease (Figure [129]4I). Immunoprecipitation (IP) assay
   with the anti‐NOTCH2 antibody revealed that adavivint enhanced the
   ubiquitination of ADAM10 (Figure [130]4J). These results suggest that
   adavivint suppresses the regulation of the Wnt pathway by specifically
   attenuating the protein expression of ADAM10 and NOTCH2.

Figure 4.

   Figure 4
   [131]Open in a new tab

   ADAM10/NOTCH2 signaling is identified as a target of adavivint to
   regulate Wnt pathway in CRCs. A) Volcano plot B) Heatmap shows the 6
   significantly changed proteins (1.5 or 0.67, P < 0.05) after adavivint
   treatment for 24 hours in P9 organoids. C) Bubble diagram shows
   relationships between RNA levels of the six genes and Wnt pathway
   enrichment scores in CRCs in the indicated datasets. D) Correlation
   analysis of enrichment scores of Wnt pathway and NOTCH pathway in CRC
   tissues (R = 0.84. P < 0.001, n = 14) and organoids (R = 0.86. P <
   0.001, n = 15). Spearman correlation. E) NOTCH receptors and NOTCH
   ligands are detected by Western blot after immunoprecipitated by
   anti‐ADAM10 body. F) Members of ADAM family are detected by Western
   blot after immunoprecipitation assay using anti‐NOTCH2 body. G) ADAM10
   and NOTCH2 are detected by Western blot after treatment of adavivint
   for 24 hours. “ADA” indicates “adavivint”. H) Wnt pathway enrichment
   analysis after knockdown of ADAM10 or NOTCH2 in the indicated organoids
   (n = 3). I) Heatmap shows relative fold change of ADAM10, NOTCH2 and
   Wnt target genes by qRT‐PCR after treated by adavivint for 24 hours. J)
   Cell lysates are analyzed by immunoprecipitation with anti‐ADAM10 and
   Western blot with indicated antibodies. The black arrow indicates the
   band location of ADAM10 protein. K) IHC analysis for ADAM10 and NOTCH2
   in representative sections from the normal tissue, primary CRC tissue,
   and liver‐metastatic CRC tissue (left), and correlation analysis
   (Spearman correlation, R = 0.86. P<0.001, n = 15) using their scores by
   IHC analysis in CRC tissue (right). Bar: 200×, 50µm. L) Histogram shows
   relative protein levels of ADAM10 and NOTCH2 in normal colorectal (n =
   4), primary (n = 6) and metastatic tumor tissues (n = 6) by IHC
   analysis. M) ADAM10 and NOTCH2 are detected by Western blot in the
   indicated organoids (Left), and Corrplot shows the relationships among
   Wnt pathway enrichment scores, relative protein levels of ADAM10 and
   NOTCH2 in CRC organoids (Right). N) After knockdown of ADAM10, ADAM10
   and NOTCH2 are detected by Western blot in the indicated organoids. All
   P‐values in (H and L) are calculated using an unpaired two‐sided
   Student's t‐test, and data presented as mean ± SD. *p < 0.05, **p <
   0.01, ***p < 0.001.

   Next, IHC was performed to analyze the expression levels of ADAM10 and
   NOTCH2 in the normal colorectal and primary and metastatic CRC tissues
   (Figure [132]4K). IHC revealed a positive association between ADAM10
   and NOTCH2 expression levels (Figure [133]4K), with the highest levels
   of ADAM10 and NOTCH2 proteins detected in the liver metastatic CRC
   tissues compared to those in the normal colorectal and primary CRC
   tissues (Figure [134]4L). We also assessed the protein levels of ADAM10
   and NOTCH2 in CRC organoids via western blotting analysis
   (Figure [135]4M). Correlation analysis revealed positive relationships
   between ADAM10, NOTCH2, and Wnt pathway enrichment scores
   (Figure [136]4M). ADAM10 knockdown decreased the protein levels of
   NOTCH2 in CRC organoids (Figure [137]4N).

2.5. ADAM10/NOTCH2 Signaling Functions as a Bypass Signaling to Regulate Wnt
Pathway by Inducing the Transcription of TCF7L2 and Wnt Target Genes in CRC

   To elucidate the potential mechanisms through which NOTCH2 regulates
   the classical Wnt pathway, we initially conducted a correlation
   analysis among the enrichment score of the Wnt pathway, AUC of
   adavivint, and the mRNAs of the four transcription factors of the
   classical Wnt pathway (TCF7, TCF7L2, TCF7L1, and LEF1). TCF7L2
   significantly correlated with the transcriptional level of the Wnt
   pathway and adavivint sensitivity (Figure [138] 5A). Of the four
   transcription factors, TCF7L2 exhibited the highest mRNA levels in CRC
   tissues and organoids (Figure [139]S4, Supporting Information). We also
   conducted Cut&Tag assays targeting NOTCH2 and TCF7L2. Of the nine
   downstream genes of the classical Wnt pathway, the results indicated
   peak enrichment of both NOTCH2 and TCF7L2 at the MYC, JUN, FOSL1,
   CCND1, CCND2, and CCND3 promoters (Figure [140]5B).

Figure 5.

   Figure 5
   [141]Open in a new tab

   ADAM10/NOTCH2 signaling directly activates transcription of TCF7L2 and
   Wnt target genes in CRC. A) Corrplot shows the relationships among AUC
   of adavivint, Wnt pathway enrichment score and mRNA level of
   transcription factor of Wnt pathway downstream in CRC organoid (n = 15,
   spearman correlation). B) In LM5R organoids, Cut&tag assay and
   sequencing analysis indicated that TCF7L2 and NOTCH2 could be enriched
   in promoters of Wnt target genes. The red bar indicates gene start
   site. C) Relative fold change of mRNA levels of the Wnt target genes
   after NOTCH2 or TCF7L2 knockdown using siRNAs in the indicated CRC
   organoids. D) Relative fold change of mRNA levels of the Wnt target
   genes after NOTCH2 or TCF7L2 stable knockdown in the indicated CRC
   organoids. E) Co‐IP assay of NOTCH2 and TCF7L2 in the indicated
   organoids. F) Relative fold change of mRNA levels of NOTCH2 or TCF7L2
   after NOTCH2 or TCF7L2 knockdown in the indicated CRC organoids (n =
   3). G) Cut&tag assay and qRT‐PCR shows that NOTCH2 directly binds to
   promoter of TCF7L2, which could be attenuated by ADAM10 knockdown (n =
   3). H,I) Western blot shows decreased TCF7L2 after ADAM10 and NOTCH2
   knockdown in the indicated CRC organoids. J) Relative mRNA levels of
   the Wnt target genes after NOTCH2 knockdown in CRC organoids
   overexpressed TCF7L2 (n = 3). K) CCK8 assay is conducted to assess
   proliferation ability of the indicated organoids treated by NOTCH2
   siRNAs (n = 3). All P‐values in (F, G, J, and K) are calculated using
   an unpaired two‐sided Student's t‐test, and data presented as mean ±
   SD. *p < 0.05, **p < 0.01, ***p < 0.001.

   To clarify the regulatory mechanisms by which TCF7L2 and NOTCH2 control
   downstream genes in the classical Wnt pathway, we initially knocked
   down TCF7L2 and NOTCH2 in CRC organoids (Figure [142]S5A,B, Supporting
   Information), which decreased the expression levels of MYC, JUN, FOSL1,
   CCND1, CCND2, and CCND3 (Figure [143]5C). Similar results were observed
   in stable knockdown CRC organoids using lentiviruses containing sgRNAs
   targeting TCF7L2 and NOTCH2 (Figure [144]S5C, D, Supporting
   Information; Figure [145]5D). However, co‐IP assay using NOTCH2 and
   TCF7L2 antibodies revealed no direct interaction between NOTCH2 and
   TCF7L2 (Figure [146]5E). A previous study also reported that the NOTCH2
   ligand, DLL1, regulates gene transcription by interacting with other
   transcriptional factors, such as TCF7L2 and Smad2/3^[ [147]^25 ^].
   However, IP assay of the TCF7L2 antibody indicated no direct interplay
   between TCF7L2 and other NOTCHs or their ligands in this study (due to
   the extremely low levels, other ligands, such as JAG2, DLL1, and DLL3,
   were not detected; Figure [148]S6, Supporting Information). NOTCH2
   knockdown decreased the TCF7L2 mRNA levels in CRC organoids
   (Figure [149]5F). Based on the results of the cut‐and‐tag and
   sequencing analyses, a pair of specific primers was designed. Cut&Tag
   and quantitative reverse transcription‐polymerase chain reaction
   (qRT‐PCR) assays revealed the direct interaction between NOTCH2 protein
   and TCF7L2 promoter, which was attenuated by ADAM10 knockdown
   (Figure [150]5G). TCF7L2 protein levels were also decreased by ADAM10
   and NOTCH2 knockdown (Figure [151]5H–I). Furthermore, TCF7L2‐mediated
   expression of MYC, JUN, FOSL1, CCND1, and CCND2 was reversed by NOTCH2
   knockdown in CRC organoids (Figure [152]5J). NOTCH2 siRNA treatment of
   CRC organoids with different WNT scores revealed the growth rate of
   organoids with high transcriptional levels of the Wnt pathway, such as
   P1, LM6, LM5R, and LM8 (Figure [153]5K). However, P4 organoids with
   lower enrichment scores for the Wnt pathway grew faster after treatment
   with NOTCH2 siRNA (Figure [154]5K). These results suggest
   ADAM10/NOTCH2/TCF7L2 as a bypass signaling pathway that regulates the
   transcription of Wnt target genes.

   Cut&Tag assays also revealed enriched peaks in the promoters of ADAM10
   and JAG1 (Figure [155]S7A, Supporting Information). qRT‐PCR (Figure
   [156]S7B, Supporting Information) and western blotting (Figure
   [157]S7C, Supporting Information) revealed that both TCF7L2 and NOTCH2
   knockdown decreased the expression of JAG1, but not ADAM10. Adavivint
   also decreased JAG1 expression in CRC organoids (Figure [158]S7D,
   Supporting Information).

2.6. Use of Adavivint and Blockage of ADAM10/NOTCH2/TCF7L2 Signaling Suppress
CRC Growth by Inhibiting Wnt Target Gene Expression

   IF analysis revealed decreased expression of ADAM10 in the cytomembrane
   and cytoplasm, as well as decreased expression of NOTCH2 in the
   cytomembrane, cytoplasm, and nucleus after 24‐h of adavivint treatment
   (Figure [159] 6A). However, the expression of β‐catenin and TCF7L2
   remained unchanged (Figure [160]6A). Compared to primary and liver
   metastatic CRC tissues, normal colorectal tissues showed lower levels
   of ADAM10 and NOTCH2 (Figure [161]6B). By nuclear cytoplasmic
   separation assay, we also observed decreased NOTCH2 in both the
   cytoplasm and nucleus, with no obvious change in the nuclear entrance
   of β‐catenin, as shown by western blotting (Figure [162]6C). Moreover,
   we did not detect MYC and JUN expression in the cytoplasm, but observed
   decreased expression in the nucleus (Figure [163]6C). Adavivint did not
   change TCF7L2 expression after 24 h of treatment, but decreased TCF7L2
   expression at 48 h (Figure [164]6D). Using the Cut&Tag and qRT‐PCR
   assays, NOTCH2 binding to the TCF7L2 promoter was attenuated by
   adavivint treatment for 24 h (Figure [165]6E). Forcing MYC and JUN
   expression rescued the adavivint‐induced growth suppression of CRC
   organoids (Figure [166]6F).

Figure 6.

   Figure 6
   [167]Open in a new tab

   Both adavivint and blockage of ADAM10/NOTCH2/TCF7L2 signaling suppress
   growth of CRC organoids by decreasing expression of Wnt target genes.
   A) IF analysis for ADAM10, NOTCH2, β‐catenin and TCF7L2 in CRC
   organoids after adavivint treatment for 24 hours (n = 4). Bar: 400×,
   50µm. B) Western blot analysis of ADAM10 and NOTCH2 in colorectal
   tissues, primary and liver‐metastatic CRC tissues. C) Western blot
   analysis of NOTCH2, β‐catenin, MYC and JUN in the indicated organoids
   after treated by adavivint (0.5 µM) for 24 hours. “ADA” indicates
   “adavivint”. D) Western blot analysis of TCF7L2 in the indicated
   organoids treated by adavivint (0.5 µM) for 24 hours and 48 hours. E)
   Cut&tag assay and qRT‐PCR shows that NOTCH2 binding to promoter of
   TCF7L2 could be attenuated by adavivint (0.5 µM) for 24 hours (n = 3).
   F) After treated by adavivint for 24 hours, CCK8 assay is performed to
   examine the proliferation ability of the indicated organoids
   transfected by MYC or JUN‐overexpressing plasmid (n = 4). G) Western
   blot analysis of MYC and JUN in the indicated organoids transfected by
   ADAM10, NOTCH2, and TCF7L2 siRNAs. H) In the indicated organoids
   transfected by ADAM10‐overexpressing plasmid for 24 hours, CCK8 assay
   is performed to assess the proliferation ability of the indicated
   organoids after transfected by MYC or JUN siRNA. I) CCK8 assay was used
   to examine the effect of NOTCH2 knockdown on proliferation ability of
   the indicated organoids overexpressing ADAM10 (n = 4). In the indicated
   organoids transfected by ADAM10 siRNA for 24 hours, J) Western blot
   analysis was used to detect MYC and JUN after transfected by MYC or
   JUN‐overexpressing plasmid, and K) CCK8 assay is further performed to
   assess the proliferation ability of the indicated organoids (n = 3).
   All P‐values in are calculated using an unpaired two‐sided Student's
   t‐test, and data presented as mean ± SD. *p < 0.05, **p < 0.01, ***p <
   0.001.

   Next, we examined the decreased protein levels of MYC and JUN in CRC
   organoids treated with ADAM10, NOTCH2, and TCF7L2 siRNAs
   (Figure [168]6G). To explore whether NOTCH2, MYC, and JUN mediate
   ADAM10‐induced organoid growth, we upregulated the expression of ADAM10
   in CRC organoids (Figure [169]S5E, Supporting Information) and observed
   the outgrowth of CRC organoids (Figure [170]6H). This outgrowth was
   reversed by MYC, JUN, and NOTCH2 knockdown (Figure [171]6H,I; Figure
   [172]S5F, G, Supporting Information). ADAM10 knockdown attenuated the
   effects of MYC and JUN overexpression (Figure [173]6J,K).

2.7. Use of Adavivint and Blockage of ADAM10/NOTCH2 Pathway Alleviate CRC
Chemoresistance by Inhibiting Wnt Target Gene Expression

   We demonstrated that adavivint inhibited CRC organoid growth by
   interfering with the Wnt pathway. Activation of the Wnt pathway is
   associated with chemoresistance in CRC.^[ [174]^26 , [175]^27 ^]
   Therefore, we investigated whether adavivint alleviates chemoresistance
   in CRC. Initially, we assessed the sensitivity of CRC organoids to the
   folinic acid, fluorouracil, and oxaliplatin (FOLFOX) regimen, which
   revealed diverse responses among different organoids (Figure [176] 7A).
   Then, we successfully generated FOLFOX‐resistant organoids using the P2
   and LM5R organoids, named P2F and LM5RF, respectively (Figure [177]7B).
   After the induction of FOLFOX resistance, enrichment score of the Wnt
   pathway increased but was reversed by adavivint in P2 organoids
   (Figure [178]7C). These two FOLFOX‐resistant organoids also exhibited
   elevated expression levels of ADAM10, NOTCH2, MYC, and JUN, but this
   trend reversed by adavivint treatment (Figure [179]7D). In the P2 and
   LM3 organoids, ADAM10 overexpression induced resistance to the FOLFOX
   regimen, which was mitigated by the knockdown of NOTCH2
   (Figure [180]7E), MYC, and JUN (Figure [181]7F). For MYC and JUN
   knockdown, we treated the CRC organoids with three siRNAs and selected
   the most effective siRNA for subsequent experiments (Figure [182]S5F,G,
   Supporting Information). By combining adavivint with the FOLFOX
   regimen, we observed a smaller synergistic index in the two
   FOLFOX‐resistant organoids than in the derived parental organoids
   (Figure [183]7G). In vivo, combined adavivint and FOLFOX regimen more
   effectively suppressed the growth of P2F and LM5RF organoids than any
   single treatment (Figure [184]7H). In the P2F and LM5RF organoids,
   ADAM10 knockdown attenuated resistance to the FOLFOX regimen
   (Figure [185]7I). The synergistic index of adavivint and FOLFOX regimen
   was also calculated for other organoids, revealing synergistic effects
   in nearly all CRC organoids, particularly in those with high ADAM10 and
   NOTCH2 levels and enrichment scores of the Wnt pathway, such as the P
   and LM5L organoids (Figure [186]7J). Based on the AUC value, we
   selected one organoid (P3 organoid) moderately sensitive to the FOLFOX
   regimen and one organoid (LM5L organoid) resistant to the FOLFOX
   regimen to establish a PDOX model. The combined adavivint and FOLFOX
   regimen most effectively suppressed the growth of the P3 and LM5L
   organoids, with stronger effects on LM5L organoids (Figure [187]7K).
   IHC analysis of the tumors showed the lowest level of Ki67 and highest
   level of caspase 3 in the combined group and moderate levels in the
   single treatment groups (Figure [188]S8, Supporting Information).
   Moreover, expression levels of ADAM10 and NOTCH2 were decreased in the
   adavivint and combined treatment groups (Figure [189]S8, Supporting
   Information).

Figure 7.

   Figure 7
   [190]Open in a new tab

   Adavivint or targeting ADAM10/NOTCH2 signaling alleviates
   chemoresistance by suppressing expression of Wnt target genes in CRCs.
   A) CCK8 assay was performed to calculate the AUC of FOLFOX in the
   indicated organoids. “Chemo” with “Y” indicated the patient had
   received regimen of FOLFOX before surgery. B) Dose‐response curves for
   regimen of FOLFOX in established FOLFOX‐resistant CRC organoids (n =
   3). C) Enrichment scores of Wnt pathway in FOLFOX‐resistant CRC
   organoids treated by adavivint (0.5 µM, n = 3). D) In FOLFOX‐resistant
   organoids, Western blot analysis is used to detect ADAM10, NOTCH2, MYC
   and JUN after treated by adavivint for 24 hours. “ADA” indicates
   “adavivint”. E) CCK8 assay is conducted to examine the survival
   inhibition ratio of FOLFOX regimen in the indicated organoids
   transfected by ADAM10‐overexpressing plasmid and NOTCH2 siRNA (n = 3).
   F) After transfected by ADAM10‐overexpressing plasmid, CCK8 assay is
   used to examine the survival inhibition ratio of regimen in the
   indicated organoids treated by MYC siRNA or JUN siRNA (n = 3). G)
   Synergistic index of adavivint and FOLFOX is calculated in the
   indicated dose in FOLFOX‐resistant organoids. H) The effect of
   combination of adavivint and FOLFOX on the growth of tumor in PDOX
   model established using the FOLFOX‐resistant organoids (n = 4). I) CCK8
   assay is examine the relative viability of the FOLFOX‐resistant
   organoids transfected by ADAM10 siRNA (n = 3). J) Heatmap shows the
   synergistic index of adavivint and FOLFOX at different concentration in
   the indicated organoids. “ADA” indicates “adavivint” K) The effect of
   combination of adavivint and FOLFOX on the growth of tumor in PDOX
   model established using the indicated organoids (n = 4). All P‐values
   in are calculated using an unpaired two‐sided Student's t‐test, and
   data presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001.

3. Discussion

   CRC is associated with highly heterogeneous and complex signaling
   networks that significantly hinder its treatment.^[ [191]^28 ^] Hence,
   identification of critical and aberrant signaling networks, whose
   blockade leads to systemic failure of cancer cells, is crucial for
   successful treatment.^[ [192]^29 ^] Tumor tissues harbor a substantial
   number of non‐tumor cells, including immune cells and fibroblasts,
   which impede the identification of key signaling pathways in
   transcriptomic analyses. Use of organoid models can circumvent this
   issue and facilitate the acquisition of high‐purity cancer cells.^[
   [193]^30 ^] Furthermore, organoid models can effectively mimic the in
   vivo growth status of tumor cells, thereby contributing to the
   identification of novel cancer driver genes and potential drug
   targets.^[ [194]^31 ^] Here, we successfully established 15 CRC
   organoids, including six primary CRC, nine liver‐metastatic CRC, and
   four colorectal organoids. CRC organoids can replicate the Wnt pathway
   activity at the transcriptional level in CRC tissues. Over 95% of
   established CRC organoids exhibit either TP53 or APC mutations.^[
   [195]^21 , [196]^23 , [197]^32 ^] Consistently, over 90% of the CRC
   organoids showed either TP53 or APC mutations in this study.

   Loss‐of‐function mutations in APC can enhance the β‐catenin/TCF
   transcriptional activity.^[ [198]^14 ^] Here, we observed higher
   transcriptional level of Wnt pathway in APC wild‐type and mutant CRC
   organoids than in the colorectal organoids. Interestingly, some CRC
   cases without APC mutation showed little expression of β‐catenin but
   high WNT scores and grew faster in the medium without RSPO1 than in the
   medium with RSPO1, indicating that the transcription of Wnt pathway is
   regulated independent of Wnt/β‐catenin signaling. Using transcriptome
   sequencing and proteomic profiling with reverse phase protein arrays,
   Menck et al. identified the Wnt pathway as a specific signature for CRC
   liver metastasis and demonstrated its association with inter‐ and
   intra‐patient heterogeneity.^[ [199]^33 ^] In this study, we also
   observed the activated transcription of the Wnt pathway in
   liver‐metastatic CRC. Targeting transcriptional regulation of Wnt
   pathway, different from targeting classical Wnt/β‐catenin/TCF, may be a
   promising therapeutic strategy to suppress CRC. Furthermore, when
   cultured in medium without RSPO1, organoids with high transcriptional
   levels of the Wnt pathway not only showed higher transcriptional
   activity of the Wnt pathway, but also activation of some pathways
   involved in tumor development, such as the Notch, Jak/Stat, and
   Hedgehog pathways.^[ [200]^34 , [201]^35 , [202]^36 ^] Maia Chanrion
   et al. reported that the Wnt pathway, Notch pathway, and TP53 mutations
   interact and maintain a balance to promote CRC progression.^[ [203]^37
   ^] Wnt pathway involves complicated regulatory networks, and targeting
   this crucial point may be a key step in the successful disruption of
   this pathway.

   To identify effective and low‐toxicity inhibitors targeting the Wnt
   pathway, we searched for known inhibitors, resulting in the acquisition
   of 24 inhibitors for subsequent drug screening. Normal colorectal
   organoids were used as negative controls. Notably, the specific targets
   of some compounds, such as adavivint and teplinovivint, are still
   unknown. Screening for these compounds may reveal pathways regulating
   the transcription of the Wnt pathway in CRC. Here, vitro and in vivo
   experiments revealed adavivint as a candidate regulating the
   transcription of the Wnt pathway. Surprisingly, its effects were not
   dependent on the classical Wnt/β‐catenin pathway, suggesting that
   adavivint disrupts transcriptional regulation of the Wnt pathway via an
   alternative mechanism. DK419, an inhibitor targeting the Wnt pathway by
   decreasing the protein levels of β‐catenin, also activates the
   transcription of the Wnt pathway. In biological processes, negative
   feedback limits the over‐activation of the Wnt pathway.^[ [204]^38 ,
   [205]^39 , [206]^40 ^] DK419 targeted this pathway by suppressing the
   core members of Wnt pathway, such as β‐catenin, but the negative
   feedback mechanism‐mediated upregulation of other members of Wnt
   pathway limited its inhibitory effects. Targeting canonical
   Wnt/β‐catenin signaling by PORCN inhibitors or RSPO3 blockage leads to
   rapid and sustained tumor regression but also induces serious side
   effects.^[ [207]^17 , [208]^18 , [209]^19 ^] Here, small‐molecule
   inhibitor tegatrabetan targeting β‐catenin showed desirable anti‐tumor
   activity but high toxicity in normal colorectal organoids. Previous
   studies reported that Wnt pathway also plays crucial roles in
   development and therapeutic resistance of LUAD and PDAC.^[ [210]^41 ,
   [211]^42 ^] In this study, adavivint suppressed the growth of LUAD and
   PDAC organoids with high enrichment scores for the Wnt pathway.
   Therefore, adavivint is a promising antitumor agent with few side
   effects for different cancer types.

   Many pathways implicated in carcinogenesis and tumor development,
   including the TP53, Notch, Sonic Hedgehog, and Jak/Stat3 pathways,
   crosstalk with the classical Wnt pathway.^[ [212]^12 , [213]^43 ,
   [214]^44 ^] Through mass spectrometry (MS) and transcriptomic analyses
   and sequential validation, we discovered that adavivint specifically
   attenuated ADAM10/NOTCH2 signaling at the protein level. Members of the
   ADAM family are multi‐domain transmembrane metalloproteases that cleave
   numerous cell surface proteins and activate downstream pathways
   involved in tumor development, including NOTCHs and their ligands.^[
   [215]^25 , [216]^45 ^] IP with anti‐ADAM10 and western immunoblotting
   with an ubiquitin antibody indicated that adavivint degrades ADAM10 via
   the ubiquitin–proteasome system.^[ [217]^46 ^] Notably, we observed
   direct interactions between ADAM10 and NOTCH2; however, no direct
   interactions were detected between ADAM10 and the other NOTCH
   receptors. Members of the ADAM family, including ADAM9, ADAM10, and
   ADAM17, mediate signaling pathways in gastrointestinal cancer
   formation,^[ [218]^47 ^] Targeting ADAM10 and ADAM17 has demonstrated
   promising antitumor activity, as exemplified by compounds such as
   INCB3619 and INCB7839.^[ [219]^48 ^] Among the three ADAMs, a direct
   interaction was observed only between NOTCH2 and ADAM10 in CRC. In the
   Apc^Min/+ model, the absence of ADAM17 resulted in a reduction in tumor
   formation in mouse intestinal cancer, highlighting the crucial role of
   ADAM17 in maintaining Wnt pathway activity.^[ [220]^49 ^] Initially,
   adavivint was discovered to block the Wnt pathway by suppressing CLK2
   and DYRK1A, and has advanced phase II clinical trials for knee
   osteoarthritis treatment.^[ [221]^22 , [222]^24 ^] In this study, we
   identified 1837 proteins using MS analysis; CLK2 and DYRK1A were not
   detected. Thus, we identified a novel target for adavivint to regulate
   transcription in the Wnt pathway. Additionally, we revealed that TCF7L2
   directly binds to the promoter of the NOTCH2 ligand JAG1 and promotes
   the transcription of JAG1, which is similar to the results of a
   previous study.^[ [223]^27 ^] Adavivint treatment also decreases JAG1
   following adavivint treatment. These results suggest a novel mechanism
   by which adavivint inhibits transcriptional regulation of the Wnt
   pathway by attenuating the expression of ADAM10 and NOTCH2 proteins.

   Crosstalk between the Wnt and Notch pathways plays a crucial role in
   maintaining stem cell‐like characteristics, as well as embryo and fetal
   development.^[ [224]^50 , [225]^51 ^] TCF7L2, a downstream
   transcription factor of the Wnt pathway, is significantly correlated
   with transcriptional activation of the Wnt pathway and sensitivity to
   adavivint. In colon carcinoma, the presence of a stable and
   constitutively active β‐catenin/TCF7L2 signaling pathway indicates the
   crucial role of TCF7L2 in carcinogenesis.^[ [226]^52 , [227]^53 ^]
   Among the four members of the TCF/LEF family, TCF7L2 exhibited the
   highest mRNA levels, indicating its pivotal role. Cut and tag assays
   using anti‐TCF7L2 and anti‐NOTCH2 antibodies revealed enriched peaks in
   the promoters of genes downstream of the Wnt pathway, including MYC,
   JUN, FOSL, CNND1, CNND2, and CNND3. Interestingly, we observed that
   NOTCH2 knockdown reversed TCF7L2‐mediated transcriptional activation of
   MYC, JUN, FOSL, CNND1, CNND2, and CNND3. We further demonstrated that
   blocking ADAM10/NOTCH2 signaling could suppress CRC organoid growth,
   involving the attenuation of transcriptional activation of Wnt
   pathways, such as MYC and JUN. Furthermore, NICD2 the intracellular
   active domain of NOTCH2) binds directly to the TCF7L2 promoter and
   promotes the transcription and translation of TCF7L2. NOTCH2 knockdown
   effectively suppresses the growth of CRC organoids at high
   transcriptional levels in the Wnt pathway. Similarly, Tian et al.
   revealed that targeting NOTUM, an extracellular palmitoleoyl‐protein
   carboxylesterase that antagonizes canonical Wnt signaling in mammals,
   is an effective approach for arresting tumor progression and metastasis
   in CRC.^[ [228]^54 ^] Additionally, we also revealed that TCF7L2 could
   directly bind to the promoter of NOTCH2 ligand JAG1 and promote
   transcription of JAG1, which was similar with previous study.^[
   [229]^27 ^] Adavivint treatment decreased JAG1 expression in CRC
   organoids. Therefore, ADAM10/NOTCH2/TCF7L2 signaling serves as a bypass
   pathway to activate the transcription of Wnt target genes, and the Wnt
   pathway can be effectively disrupted by blocking this signaling
   pathway.

   In CRC, activation of the Wnt and Notch pathways is associated with
   chemotherapy resistance, including resistance to 5‐FU and
   oxaliplatin.^[ [230]^55 , [231]^56 , [232]^57 ^] Among the downstream
   genes of the canonical Wnt pathway, we observed that
   ADAM10/NOTCH2/TCF7L2 signaling regulates MYC and JUN more prominently
   than the other genes. MYC and JUN, well‐known proto‐oncogenes, promote
   tumor growth and induce drug resistance.^[ [233]^58 , [234]^59 ,
   [235]^60 ^] Our study revealed that ADAM10/NOTCH2 signaling promotes
   tumor growth and chemotherapy resistance by activating the
   transcription of MYC and JUN in CRC. In triple‐negative breast cancer,
   ADAM10 is implicated in oncogenic processes and chemoresistance by
   regulating the NOTCH1 signaling pathway.^[ [236]^61 ^] Many proteins
   were revealed as the substrates of ADAM10, such as NOTCH ligands
   (Dll1), Amyloid precursor protein (APP) and E‐cadherin, etc.^[ [237]^62
   ^] Our results indicated that ADAM10 may tend to bind NOTCH2 in the
   context of colorectal organoids, and ADAM10 may interact with other
   NOTCHs in other types of cells or at certain conditions. The potential
   mechanisms are still to be explored in the further studies. In human
   breast cancer cells, the ADAM10 and ADAM17 inhibitor, INCB3619, along
   with a lapatinib‐like dual inhibitor of endothelial growth factor
   receptor and HER‐2/neu kinases, demonstrated synergistic growth
   inhibition.^[ [238]^48 ^] Aderbasib, another inhibitor of ADAM10 and
   ADAM17, overcomes endothelial growth factor receptor inhibitor
   resistance and enhance fluorouracil sensitivity in CRC.^[ [239]^63 ^]
   Simultaneously, using a CRC organoid model in vitro and in vivo, we
   observed that adavivint enhanced FOLFOX sensitivity by suppressing
   ADAM10/NOTCH2 signaling‐mediated regulation of Wnt target genes.
   Therefore, transcriptional activation of the Wnt pathway and high
   levels of ADAM10/NOTCH signaling are indications for adavivint
   treatment of CRC.

   In conclusion, this study conducted integrated transcriptomic analysis
   of tumor organoids and tissues to assess the transcriptional regulation
   of the Wnt pathway and its association with liver metastasis in CRC. By
   screening inhibitors targeting the Wnt pathway in vitro and in vivo, we
   identified adavivint as a promising anticancer agent for CRC treatment.
   Mechanistically, adavivint suppressed ADAM10/NOTCH2/TCF7L2 signaling,
   which served as a bypass pathway to activate the transcription of Wnt
   target genes. Adavivint treatment and blockage of ADAM10/NOTCH2/TCF7L2
   signaling effectively induced growth arrest and susceptibility to
   chemotherapy (Figure [240] 8 ). Overall, this study provides a
   promising anticancer agent and suggests a potential target for CRC
   treatment.

Figure 8.

   Figure 8
   [241]Open in a new tab

   Mechanism of adavivint suppressing bypass ADAM10/NOTCH2/TCF7L2
   signaling mediated transcriptional regulation of Wnt pathway to
   suppress tumor growth and overcome chemoresistance in CRC.

4. Experimental Section

Organoid Culture

   Eight primary CRC and paired normal colorectal tissues were obtained
   from the Shanghai East Hospital of Shanghai Tongji University, which
   was approved by the Institutional Review Board of the Shanghai East
   Hospital of Shanghai Tongji University. Eight CRC metastatic and four
   paired normal liver tissues were obtained from the Department of
   Hepatic Surgery, Fudan University Shanghai Cancer Center, which was
   approved by the Institutional Review Board of the Fudan University
   Shanghai Cancer Center. Two primary LUAD and two PDAC tissues were
   obtained from the Department of Thoracic Surgery, Ruijin Hospital of
   Shanghai Jiaotong University School of Medicine and Department of
   Hepatobiliary Surgery, Shanghai Sixth People's Hospital of Shanghai
   Jiaotong University School of Medicine, respectively. This study was
   approved by the Institutional Review Board of Ruijin Hospital of
   Shanghai Jiaotong University School of Medicine and Shanghai Sixth
   People's Hospital of Shanghai Jiaotong University School of Medicine.
   Organoid culture was performed as previously described.^[ [242]^64 ^]
   Written informed consent was obtained from all patients before tissue
   acquisition. Clinical characteristics of the patients were presented in
   Table [243]S3 (Supporting Information). Culture media used for CRC
   organoids (WM‐H‐03), PDAC organoids (WM‐H‐05), LUAD organoids
   (WM‐H‐10), human colorectal organoid expansion (WM‐H‐01), and human
   colorectal differentiation (WM‐H‐01D) were purchased from Shanghai
   OuMel Biotechnology Co., Ltd. To explore whether CRC organoids grow
   independent of RSPO1, another organoid culture medium was prepared with
   advanced Dulbecco's modified Eagle's medium (DMEM)/F12 (Gbico), 100
   µg mL^−1 human RSPO1 (Sino Biological), 100 µg mL^−1 human noggin (Sino
   Biological), 1 × B27 (Gbico), 1.25 mM n‐acetyl cysteine (MCE), 10 mM
   nicotinamide (MCE), 60 ng/mL human EGF (Sino Biological), 100 ng mL^−1
   human FGF10 (Sino Biological), 10 nM gastrin (MCE), 500 nM A83‐01
   (MCE), 3 µM SB202190 (MCE), 20 nM prostaglandin E2 (MCE), and 100
   µg mL^−1 primocin (InvivoGen).

Whole Exome Sequencing Analysis

   After end‐repairing, A‐tailing, and ligation, the fragmented genomic
   DNA was sequentially mixed with indexed adapters, followed by size
   selection using the Agencourt AMPure XP beads (Beckman Coulter Inc.,
   Brea, CA, USA). Library was constructed using the KAPA Library
   Preparation kit (Kapa Biosystems, Inc., Wilmington, MA, USA), and PCR
   (7–8 cycles according to the amount of DNA used) was performed with
   pre‐capture ligation‐mediated PCR (Pre‐LM PCR) Oligos (Kapa Biosystems,
   Inc.) in 50‐µL reaction mixtures. DNA sequencing was performed on an
   Illumina NovaSeq 6000 system, according to the manufacturer's
   instructions, at an average depth of 300×. Sequencing data were aligned
   to the hg38 genome (GRch38) using the Burrows‐Wheeler Aligner. SAMtools
   was used to sort the BAM files and perform duplicate markings. After
   removing the duplicate reads using Gencore version 0.12.0
   ([244]https://github.com/OpenGene/gencore), single nucleotide variants
   and insertions/deletions were detected in tumor and matched‐normal
   pairs using Mutect2 from GATK. MutSigCV (version 1.41) was used to
   determine the significantly mutated genes with q‐value < 0.05.

RNA‐Seq Analysis

   Fresh tissues (normal colorectal tissues: n = 7; primary CRC tissues: n
   = 7; normal liver tissue: n = 4; liver‐metastatic CRC tissues: n = 9)
   and cultured organoids (colorectal organoid: n = 4; primary CRC
   organoids: n = 6; liver‐metastatic organoids: n = 9) were used for
   RNA‐seq. After treatment with adavivint for 24 h or siRNA for 96 h, the
   organoids were harvested for RNA‐seq. Library construction was
   performed using the TruSeq RNA Sample Preparation Kit (Illumina), and
   RNA sequencing was performed using Illumina HiSeq Xten (Illumina).
   After quality control, the gene counts were normalized to transcripts
   per million (TPM) and log2(TPM + 1)‐transformed expression data for
   further analysis. Finally, 189 gene sets of KEGG pathways were
   downloaded from GSEA database
   ([245]https://www.gsea‐msigdb.org/gsea/index.jsp), and enrichment
   scores of these gene sets were calculated using the “GSVA” package in R
   software. GSVA, a single‐sample gene set enrichment method, was
   performed to estimate the KEGG pathway activity by transforming an
   input gene‐by‐sample expression data matrix into a corresponding
   gene‐set‐by‐sample expression data matrix. The resulting expression
   data matrix was used with classical analytical methods to assess the
   differential expression between normal and tumor groups. After
   calculating the enrichment scores of KEGG pathways, difference analysis
   was performed between two groups. All RNA‐seq data used in this study
   were shown in [246]File S1 (Supporting Information).

Transfection of Organoids

   Serum‐free DMEM containing 2% basement membrane component (OM‐MG‐02;
   OuMel Biotechnology Co., Ltd.) was plated in a 24‐well plate (400
   µL well^−1) and incubated at 37 °C for 2 h. Organoids were dispersed in
   a dissociation solution (OM‐D‐01; OuMel Biotechnology Co., Ltd.), and
   the number of dissociated cells was calculated. After removing the
   medium from the 24‐well plate, the organoid medium containing 10^5
   cells was plated in a 24‐well plate for subsequent experiments. On the
   second day, the medium was removed and 500 µL serum‐free DMEM
   containing siRNA or plasmid wrapped by Lipofectamine 2000 (Invitrogen,
   Carlsbad, CA, USA) was added. After culturing for 24 h, the medium was
   exchanged with the organoid culture. The plasmids and siRNAs used in
   this study were purchased from GeneChem (Shanghai, China) and
   GenePharma (Shanghai, China), respectively. All siRNA sequences were
   listed in Table [247]S4 (Supporting Information). The lentivirus
   containing sgRNA targeting TCF7L2 or NOTCH2 was purchased from
   GeneChem. All sgRNA sequences were listed in Table [248]S4 (Supporting
   Information). The most effective siRNA or sgRNA targeting TCF7L2 or
   NOTCH2 and the most effective sequences were used for subsequent
   experiments.

High‐Throughput Screening of Wnt Pathway‐Targeting Inhibitors Using CRC
Organoids

   Organoids in good conditions (passages 3–8) were dispersed in a
   dissociation solution. Then, 4 µL mixture of basement membrane
   components and DMEM (1:1) containing 2 × 10^3 cancer cells was placed
   in every well of a 96‐well plate. On the third day, the 1 µM
   concentrations of compounds were added, and the organoids were cultured
   for another four days. For the FOLFOX regimen, the concentration ratio
   of 5‐fluorouracil, calcium folinate, and oxaliplatin was 25:5:1. On the
   seventh day, the medium was removed, and a fresh medium containing 10%
   cell counting kit‐8 was added. After incubation for 3 h, the absorbance
   (OD) was measured at 450 nm by enzyme marker.

Cell Proliferation Assay

   Cells transfected with organoids were dispersed in a dissociation
   solution (OM‐D‐01; OuMel Biotechnology Co., Ltd.). Then, 4 µL mixture
   of basement membrane components (OM‐MG‐02; OuMel Biotechnology Co.,
   Ltd.) and DMEM (1:1) containing 2 × 10^3 cancer cells was placed in
   every well of a 96‐well plate. On the first, third, and fifth days,
   absorbance (OD) values were measured at 450 nm using the cell counting
   kit‐8 kit was purchased from (DOJINDO, CK04, Kumamoto, Japan). To
   assess and quantify the morphological parameters, brightfield images
   from four independent wells were captured (40×), and organoid number
   and mean organoid area (π× radius [µm] × radius [µm]) were calculated
   as previously described.^[ [249]^65 ^]

Western Blotting and qRT‐PCR

   As the organoids grew to the desired number, total protein and total
   RNA were extracted using the TRIzol solution (15596‐026; Invitrogen)
   and lysis buffer (P0013B; Beyotime, Beijing, China), and western
   blotting and qRT‐PCR assays were performed as previously described.^[
   [250]^64 ^] The working dilutions of antibodies used here were listed
   in Table [251]S5 (Supporting Information). Reverse transcription kit
   (FSQ‐101; TOYOBO, Osaka, Japan) and SYBR Select Master Mix (4472908;
   Applied Biosystems, Foster City, USA) were used for qRT‐PCR in a 10‐µL
   reaction mixture with HT 7900 (Applied Biosystems). All primer
   sequences were listed in Table [252]S6 (Supporting Information).

HE Staining, IHC, and Immunofluorescence (IF) Analysis

   Tissue sections were prepared, and HE staining, IHC, and IF analysis
   were conducted as described in this previous report.^[ [253]^64 ^] All
   working dilutions of antibodies used here were listed in Table [254]S5
   (Supporting Information). IHC was used to quantify the expression
   levels of ADAM10 and NOTCH2 as described in this previous study.^[
   [255]^64 ^] For IF experiments, the mean fluorescence intensity of
   triplicate experiments was calculated using the ImageJ software.

IP Assay

   As the organoids grew to the desired number, basement membrane
   components were removed by adding phosphate‐buffered saline containing
   10 mM EDTA. Organoids were harvested using an IP lysis buffer (P0013;
   Beyotime). Antibodies against ADAM10, TCF7L2, and TCF7L2 and magnetic
   beads (20 µL; 88802; Thermo Scientific, MS, USA) were added to the
   lysate and incubated overnight at 4 °C. After washing thrice with the
   lysis buffer, western blotting was performed to analyze the
   precipitates. The antibody concentrations were presented in Table
   [256]S5 (Supporting Information).

Proteome Analysis Using Liquid Chromatography‐Tandem MS (LC‐MS/MS)

   After 24‐h adavivint treatment, the total protein was extracted and
   quantified via high‐resolution MS using the MaxQuant and Perseus
   software to screen the differentially expressed proteins and
   subsequently perform bioinformatic analysis. Briefly, 50 µg of protein
   was mixed with 50 mM ammonium bicarbonate solution to make up the
   volume to 50 µL. Then, dithiothreitol was added to make the final
   concentration 10 mM and incubated at 37 °C for 60 min. Iodoacetamide
   was added to a final concentration of 50 mM and allowed to react for 30
   min in the dark at room temperature. The solution was transferred to an
   ultrafiltration tube for high‐speed centrifugation and washed twice
   with 50 mM ammonium bicarbonate (containing 0.8% SDC). Trypsin was
   added and hydrolyzed in a 37 °C incubator for 12–16 h. The enzymatic
   hydrolysates were washed and pooled with 50 mM ammonium bicarbonate.
   TFA was added for acidification, and the mixture was centrifuged to
   remove SDC and desalted using the C18 desalting column. Finally, the
   desalted samples were lyophilized and resolved using 0.1% FA for MS
   analysis.

Cleavage Under Targets and Tagmentation (Cut&Tag) Assay for RNA‐Seq and PCR
Analyses

   Organoids with good growth were digested to obtain a single cell
   suspension. After counting the cells, 5 × 10^5 cells were used for the
   subsequent assay. Cut&tag assay was performed using the Hyperactive
   Universal CUT & Tag Assay Kit for Illumina (TD903; Vazyme, Nanjing),
   according to the manufacturer's instructions. Sequencing libraries were
   prepared using the TruePrep Index Kit V2 (TD202; Vazyme), according to
   the manufacturer's instructions. The PCR products were cleaned using
   clean VAHTS DNA beads (N411‐01‐AA; Vazyme). RNA sequencing was
   performed using Illumina HiSeq Xten (Illumina) at the HaploX Genomics
   Center (Shanghai, Jiangxi, China). The Trim_galore package was used to
   remove the adaptors, and the trimmed reads were aligned to the hg38
   reference genome using Bowtie2. Narrow peaks were called using the
   MACS2 package, and IGV files were produced using bamCoverage of
   deepTools for visualization in IGV 2.16.0. Finally, peak annotation and
   motif finding were performed using the Homer software. Simultaneously,
   the immunoprecipitated DNA was harvested and analyzed using PCR.
   Specific primers for TCF7L2 promoter designed according to peak calling
   and annotation analysis were listed in Table [257]S6 (Supporting
   Information).

Establishment of FOLFOX‐Resistant Organoids

   FOLFOX‐sensitive organoids (P2 and LM5R) were selected to construct the
   drug‐resistant strains. When the growth was optimum, these organoids
   were treated with FOLFOX containing 10 µM 5‐fluorouracil for two days
   (concentration ratio of 5‐fluorouracil, calcium folinate, and
   oxaliplatin = 25:5:1). The culture medium was replaced every two days
   until the organoids reached a certain count. Then, these organoids were
   treated with a low‐dose FOLFOX regimen (1 µM 5‐fluorouracil). When the
   organoids reached a certain count, the sensitivity to FOLFOX was
   tested. If the IC50 of the resistant organoids was five‐times greater
   than that of the parental organoids, the establishment of
   FOLFOX‐resistant organoids was considered successful.

PDOX Model

   Animal experiments were approved by the Institutional Animal Care and
   Use Committee of Fudan University (FUSCC‐IACUC‐S20210285). As the
   organoids grew to the desired number, the dome‐containing organoids and
   organoid medium were fully mixed and transferred to centrifugal tubes.
   After centrifugation (800 rpm) for 5 min, the medium was removed, and
   the organoids were re‐suspended in serum‐free DMEM. To count the cells
   of organoids per mL, 50 µL medium containing organoids was used for
   dissociation and calculation. Then, 100 µL mixture of organoids
   (5 × 10^5) and basement membrane components (1:1) was inoculated
   subcutaneously. When the subcutaneous tumor grew to 3–4 mm in diameter,
   drugs were administered. Adavivint (5 mg/kg or 50 mg/kg) was
   intraperitoneally injected daily, and FOLFOX (5‐fluorouracil/calcium
   folinate/oxaliplatin = 50/39/6 mg/kg) was intraperitoneally
   administered twice in one week. The tumor size was measured once a week
   and calculated as width × width × length/2.

Data Extraction and Analysis

   The gene count data of 635 CRC cases and APC mutation data of 533 CRC
   cases was downloaded from TCGA database
   ([258]http://cancergenome.nih.gov/). Gene counts were normalized to
   TPM, and log2(TPM + 1)‐transformed expression data were used for
   further analysis. Finally, 189 gene sets of KEGG pathways were
   downloaded from GSEA database
   ([259]https://www.gsea‐msigdb.org/gsea/index.jsp), and enrichment
   scores of these gene sets were calculated using the “GSVA” package in R
   software.

Statistical Analyses

   Bubble diagram was plotted by using “ggplot2” and “ggsave” package in R
   software. Student's t‐test was performed using the GraphPad Prism 6.0
   software. A histogram was used to observe the‐Log10P values determined
   in Gene Ontology analysis. Heatmap of spearman correlation values of
   organoids based on RNA‐seq data using 10000 most variable genes were
   plotted using the “ggplot2” and “data.table” packages in R software.
   Pearson correlation between two variables was analyzed using the
   “corrplot” and “PerformanceAnalytics” packages in R software or
   GraphPad Prism 6.0. Heatmap and Volcano plot were plotted using the
   “pheatmap” and “ggplot2” packages in R software, respectively. Gene
   mutation and copy number were determined using the “ComplexHeatmap”
   package in R. Tumor growth curves and dose–response curves were
   plotted, and AUC values was calculated using GraphPad Prism 6.0.
   Statistical significance was set at P< 0.05. All results were
   represented as the mean ± standard deviation of three experiments.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   L.W., G.X., Y.Z., and W.X. conceived the project. Z.X., Y.Z., Y.W., and
   X.M. designed the experiments. Z.X., Y.W., L.F., B.Y., and Y.W.
   conducted experiments, such as Cut&Tag, RNA‐seq, IHC, Western blot,
   CoIP, etc. Z.X., S.S., and J.Z. analyzed the data. Z.X., S.Y. and Y.H.
   wrote the manuscript. Y.Z. and G.X. reviewed and polished the
   manuscript.

Supporting information

   Supporting Information
   [260]ADVS-12-2405758-s001.docx^ (3.1MB, docx)

   Supplementary File S1
   [261]ADVS-12-2405758-s002.xlsx^ (28.2MB, xlsx)

Acknowledgements