ARTICLE https://doi.org/10.1038/s41467-022-30382-9 OPEN p53 wild-type colorectal cancer cells that express a fetal gene signature are associated with metastasis and poor prognosis Laura Solé 1,2,9, Teresa Lobo-Jarne1,2,9, Daniel Álvarez-Villanueva 1,3, Josune Alonso-Marañón1, Yolanda Guillén1, Marta Guix 4, Irene Sangrador1, Catalina Rozalén1, Anna Vert1, Antonio Barbachano 2,5, Joan Lop6, Marta Salido6, Beatriz Bellosillo 2,6, Raquel García-Romero1, Marta Garrido1, Jessica González1, 1234567890():,; María Martínez-Iniesta3, Erika López-Arribillaga 1,7, Ramón Salazar2,3, Clara Montagut2,4, Ferrán Torres 8, Mar Iglesias2,6, Toni Celià-Terrassa 1, Alberto Muñoz2,5, Alberto Villanueva 3, Anna Bigas 1,2 ✉ & Lluís Espinosa 1,2 ✉ Current therapy against colorectal cancer (CRC) is based on DNA-damaging agents that remain ineffective in a proportion of patients. Whether and how non-curative DNA damage- based treatment affects tumor cell behavior and patient outcome is primarily unstudied. Using CRC patient-derived organoids (PDO)s, we show that sublethal doses of chemotherapy (CT) does not select previously resistant tumor populations but induces a quiescent state specifically to TP53 wildtype (WT) cancer cells, which is linked to the acquisition of a YAP1- dependent fetal phenotype. Cells displaying this phenotype exhibit high tumor-initiating and metastatic activity. Nuclear YAP1 and fetal traits are present in a proportion of tumors at diagnosis and predict poor prognosis in patients carrying TP53 WT CRC tumors. We provide data indicating the higher efficacy of CT together with YAP1 inhibitors for eradication of therapy resistant TP53 WT cancer cells. Together these results identify fetal conversion as a useful biomarker for patient prognosis and therapy prescription. 1 Cancer Research Program, Institut Mar d’Investigacions Mèdiques, CIBERONC, Hospital del Mar, Barcelona 08003, Spain. 2 CIBERONC, Instituto de Salud Carlos III, Madrid, Spain. 3 Chemoresistance and Predictive Factors Group, Program Against Cancer Therapeutic Resistance (ProCURE), Catalan Institute of Oncology (ICO), Oncobell Program, Bellvitge Biomedical Research Institute (IDIBELL), L’Hospitalet del Llobregat, Barcelona, Spain. 4 Department of Oncology, Institut Mar d’Investigacions Mèdiques, CIBERONC, Universitat Pompeu Fabra, Barcelona 08003, Spain. 5 Department of Cancer Biology, Instituto de Investigaciones Biomédicas ‘Alberto Sols’, Spanish National Research Council (CSIC)-Autonomous University of Madrid (UAM) and IdiPAZ, Madrid, Spain. 6 Department of Pathology, Institut Mar d’Investigacions Mèdiques, Universitat Autònoma de Barcelona, Barcelona 08003, Spain. 7 Group of Biomedical Genomics, Institut de Recerca Biomedica (IRB), Barcelona 08028, Spain. 8 Biostatistics Unit, Medical School, Universitat Autònoma de Barcelona, Barcelona, Spain. 9These authors contributed equally: Laura Solé, Teresa Lobo-Jarne. ✉email: abigas@imim.es; lespinosa@imim.es NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 C olorectal cancer (CRC) remains the second leading cause concentrations led to the eradication of most PDO cells inde- of cancer-related death, which highlights the need for pendently of their mutational status. However, we were able to novel therapies focused on the treatment of advanced define in all cases the IC20 and IC30 as the 5-FU + Iri. doses that disease. Treatment of localized CRC currently involves surgery, reduced cell viability by 20 and 30% after 72 h of treatment, which radiotherapy, and/or chemotherapy (CT) (mainly 5-FU or cape- was specific for each PDO (Fig. 1A and Supplementary Table S1). citabine and oxaliplatin in the neoadjuvant or adjuvant setting), Microscopy analysis of PDO5 (TP53 WT) treated at IC20 and IC30 while CT (5-FU, oxaliplatin, and irinotecan) still represents the did not reveal obvious signs of cell death, but we observed a dose- main backbone of treatment for advanced CRC. In general, dependent growth arrest in all tested PDO TP53 WT and the classical CT agents are designed to eradicate tumors by inducing hypomorphic PDO4 that continued for at least 2 weeks after drug DNA damage in highly proliferative cells leading to cell death washout (Fig. 1B, C and S1A). In contrast, the TP53 deficient (reviewed in ref. 1). However, most tumors contain a variable PDO8, PDO10, PDO11, and PDO15 totally failed to reinitiate proportion of quiescent cells, including cancer stem cells, that are tumor growth after drug washout (Fig. S1A). Growth arrest in refractory to these agents thus contributing to tumor relapse and TP53 WT PDOs was associated with proliferation inhibition as metastasis2. Supporting this notion, the presence of intestinal determined by immunohistochemistry (IHC) analysis of ki67 stem cell (ISC) signatures in tumors is predictive of poor prog- (Fig. 1D and S1B) and reduced number of cells in S phase with nosis in patients3. In addition, although current therapeutic accumulation in G0/G1 and G2/M (Fig. S1C), the latter probably regimes are devised to expose tumor cells to the maximum doses corresponding to cells not undergoing cytokinesis17,18. By fluor- tolerated by patients, it was demonstrated that altered vascular- escent in situ hybridization (FISH) and DAPI staining, we ization in tumors led to heterogeneous drug delivery thus demonstrated the absence of polyploid or multinucleated cells impairing the efficacy of CT4,5. Consequently, even after adequate following IC30 treatment (Fig. S1D). We determined whether IC20 treatment, ~25–30% of CRC patients in the less aggressive stage II and IC30 treatments inflicted a senescent phenotype to the TP53 tumors and up to 30–50% in stage III relapse, and most of them WT PDO5 cells by evaluation of senescence-associated (SA)-β- eventually die (data from the American Cancer Society). Galactosidase activity by flow cytometry (Fig. 1E) and IHC To avoid therapeutic resistance, strategies potentiating the (Fig. S1E). Cells that persisted after IC20 or IC30 were not effect of DNA-damaging agents are recurrently proposed as the senescent, in contrast with cells treated at IC60 for 72 h (Fig. 1E). base for more effective combination therapies6,7. Sublethal CT Accordingly, the addition of the senolytic agent dasatinib19 did has also been suggested as an alternative treatment based on its not potentiate the growth inhibition imposed by IC20 and IC30 5- ability to impose a senescent phenotype on cancer cells, char- FU + Iri. but enhanced the effect of IC60 treatment (Fig. 1F). acterized by high levels of the cell cycle inhibitors p16 and p21, Moreover, we did not detect apoptotic cells in PDO5 after IC30 cessation of proliferation, and presence of a senescent-associated- treatment as determined by cleaved-caspase 3 (cCas3) staining secretory-phenotype, which can delay disease progression8 but (Fig. S1F) and Annexin V staining (Fig. S1G) that was robustly also provide pro-tumorigenic factors to neighboring tumor detected in the TP53 mutant PDO4 and PDO8 (Fig. S1H). Cell populations (reviewed in refs. 9–11). cycle arrest in the absence of apoptosis was observed not only in Recent studies in glioblastoma and squamous cell carcinoma PDO5 but also in the TP53 WT PDO20 and PDO66 (Fig. S1I). indicated that TGFβ or kinase inhibitor therapies can increase We studied whether cell cycle arrest after IC20 and IC30 treatment drug resistance by imposing a reversible-quiescent state on cancer was linked to sustained DNA damage. Comet assay (Fig. 1G) and cells12,13. Similarly, using CRC xenograft models, it was demon- Western Blot (WB) analysis of γH2A.X (Fig. 1H) in PDO5 strated that tumor cells that persist after CT display a phenotype revealed a dose-dependent accumulation of DNA damage starting resembling the quiescent/slow-cycling embryonic diapause-like at 1–3 h with a maximum at 24 h. DNA damage was undetectable stage14. YAP1 pathway, which drives fetal conversion in the at 72 h after IC20 and IC30 treatment, but clearly present in IC60- regenerating intestine, has consistently been recognized as a treated PDO5 (Fig. 1H, I). In contrast, PDO4 and PDO8 cells tumor promoter and inducer of CT resistance in cancer15. carrying mutated TP53 exhibited high amounts of DNA damage However, recent work (using the Apc−/−; KrasG12D; p53−/− following IC20 and IC30 5-FU + Iri. treatment that lasted for at murine CRC model) indicated that fetal reprogramming induced least 72 h (Fig. 1J). Further supporting the differential response to by YAP1 led to tumor and metastasis suppression16. Thus, to date sublethal CT in terms of DNA repair according to TP53 status, we there is no conclusive data to establish the impact of YAP1 and detected massive amounts of γH2A.X by WB analysis of TP53 fetal/embryonic conversion in CRC, and whether sublethal CT mutant PDOs and CRC cell lines upon IC20 treatment (Fig. S1J can directly impose specific adjustments to cancer cells. and S1K). We here show that CRC patient-derived organoids (PDO)s and In most TP53 mutant PDO cells included in our study, IC20-30 cell lines exposed to sublethal CT acquire a non-senescent usually requires a higher concentration of 5-FU + Iri., which quiescent-like phenotype that persists after DNA damage reso- could contribute to the higher DNA damage that is produced. To lution, which is restricted to TP53 WT cells. Cells acquiring this test this possibility, we used two different paired models with persistent quiescent-like (PQL) phenotype display higher in vivo comparable sensitivity to 5-FU + Iri. (see Table S1); PDO5 WT metastatic capacity and express a YAP1-dependent fetal signature and TP53 KO (generated by CRISPR-Cas9) (Fig. 1K, L) and the that is also detected in a subset of untreated CRC tumors. The HCT116 (TP53 WT) and DLD1 CRC (TP53 mutated) cell lines presence of this specific fetal signature, or detection of nuclear (Fig. 1M). We confirmed a DNA repair defective phenotype in YAP1 in tumors, predicts poor disease outcomes at stages II and p53 deficient cells treated at the same 5-FU + Iri. doses. III in patients carrying tumors with WT TP53. These results indicate that cancer cells carrying functional p53 respond to low CT by acquiring a quiescent-like phenotype, hereafter referred to as PQL, in the absence of sustained DNA Results damage. Low-dose CT treatment induces a non-senescent quiescent-like phenotype to TP53 WT cancer cells in the absence of sustained DNA damage. To investigate the mechanisms imposing therapy PQL cancer cells display increased in vitro and in vivo tumor resistance in cancer patients, we treated CRC PDOs with serial initiation capacity. We studied whether PQL cells preserved dilutions of CT agents 5-FU + Iri. As expected, high 5-FU + Iri. comparable tumor-initiating capacity (TIC) as untreated cancer 2 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE A B C 800 Untreated PDO diameter (μm) 5-FU+Iri. IC20 Lumin. (RLU*106/s) 100 8 Lumin. (RLU*106/s) 5-FU+Iri. IC30 600 75 6 **** **** 50 4 400 **** Untreated 2 **** IC20 IC30 25 **** 5-FU+Iri. IC20 200 n.s. 5-FU+Iri. IC30 **** **** 0 0 **** **** -2 -1 0 1 2 3 0 1 2 0 log[5-FU+Iri](μg/ml) Washout (weeks) 0 1 2 Washout (weeks) D E PDO5 (p53 WT) Untreated 5-FU+Iri. IC20 Untreated 5-FU+Iri IC20 5-FU+Iri IC30 10 5 2.94±1.33 5.47±0.96 p=0.056 4 10 > IF: ki67/DAPI 3 10 SA-β-GAL (FITC) -3 10 5-FU+Iri. IC30 5-FU+Iri. IC60 50 μm 50 μm 50 μm 5 10 9.37±3.57 18.8±9.42 p=0.043 p=0.044 %Ki67+ cells/sphere 4 120 **** 10 100 **** **** 3 10 80 60 40 10 -3 20 0 60K 120K 0 60K 120K 0 > FSC-A r. 20 30 nt IC IC U F G PDO5 H Relative growth (%) **** IC60 IC30 IC20 Cont. 100 **** 5-FU+Iri. **** **** IC20 (h) 0 1 3 6 24 48 72 75 120 p=0.0372 (kDa) Tail moment 100 WB: 50 15 80 γH2A.X Untreated n.s. 60 IC20 0.033 Densit. 1 1.6 1.5 2.3 4.7 2.7 1.8 25 IC30 40 IC60 20 0 histone H3 15 0 IC . -2 -1 0 1 2 3 60 r 20 30 nt IC U IC log[dasatinib](μg/ml) 3h treatment I PDO5 J M γH2A.X/DAPI **** IC60 IC30 IC20 Cont. **** PDO4 PDO8 Untreated 5-FU+Iri. IC30 n.s. **** PDO8 IC30 IC20 Cont. n.s. n.s. p=0.0004 HCT116 (p53 WT) 120 120 120 Tail moment Tail moment Tail moment **** **** 100 100 **** 100 **** 80 80 80 60 60 60 40 40 40 20 20 20 0 0 0 IC . IC . IC . r 20 r 30 20 30 60 r 20 30 nt nt nt U IC U IC IC U IC 50 μm 50 μm 72h treatment 72h treatment PDO5 DLD1 (p53 MUT) K PDO5 L p=0.0296 5-FU+Iri. p53 WT p53 KO n.s. **** n.s. (hours) 0 3 6 24 48 72 0 3 6 24 48 72 150 n.s. Tail moment (kDa) WB: 100 15 γH2A.X 50 50 μm 50 μm Histone H3 15 0 p=0.0021 25 h. h. 3 r. 3 r. 72 . 72 . p=0.0009 h h n.s. nt nt % γH2AX+ cells U U p53 50 20 p53 WT p53 KO 15 10 5 0 h. h. r. r. nt nt 72 72 U U HCT116 DLD1 cells. We seeded 300 single cells from untreated or 5-FU + Iri. panels), consistent with their low proliferation rates. In contrast, IC20 or IC30-treated TP53 WT PDO5 and PDO66. We found that 5-FU + Iri. pre-treatment (IC20) of TP53 mutant PDO4 and low-dose CT treatment of PDO5, PDO20, and PDO66 cells did PDO8 cells resulted in TIC abrogation (Fig. 2B). Similar results not preclude TIC in vitro, as indicated by the slight reduction in were obtained by KO of TP53 in PDO5 (Fig. 2C). Considering the number of spheres generated compared to controls (Fig. 2A that TIC activity in TP53 WT PDOs could be driven by the and S2A, upper panels), but imposed a reduction of spheres fraction of cells that still undergo replication after IC20 and IC30 diameter, more pronounced in PDO5 (Fig. 2A and S2A, lower treatment (Fig. 1D, S1B and S1C), we next compared the TIC NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 Fig. 1 Low-dose CT treatment induces a non-senescent quiescent-like state to CRC PDO in the absence of persistent DNA damage. A Dose-response assay of PDO5 treated with 5-FU + Iri. for 72 h, indicating the IC20 and IC30 doses (n = 3 replicates examined, from one out of four biologically independent experiments). B, C Quantification of PDO5 B viability (n = 3 replicates examined, from 1 out of 4 biologically independent experiments) and C diameter (n = at least 15 spheres examined whenever possible over four biologically independent experiments), after 72 h of 5-FU + Iri. treatment and 1–2 weeks of washout. D Representative ki67 stainings of PDO5 treated for 72 h as indicated, and quantification of ki67+ cells/sphere (n = 25 spheres examined over 3 biologically independent experiments). E Flow cytometry analysis of SA-β-Gal activity in PDO5 treated as in D. F Dose–response curves of PDO5 treated with dasatinib for 3 days after 5-FU + Iri. pre-treatment, as indicated (n = 2–3 biologically independent experiments). G Comet assay in PDO5 treated for 3 h as indicated (n = more than 700 cells examined over three independent experiments). H WB analysis of the DNA damage sensor γH2A.X in PDO5 cells collected at the indicated time points after 5-FU + Iri treatment (from one out of three biologically independent experiments). I, J Comet assay in 53 WT PDO5 I and p53 mutants PDO4 and PDO8 J, treated for 72 h as indicated (n = >800 cells examined over 3 independent experiments). K, L WB analysis K (from one out of three biologically independent experiments) and comet assay L of PDO5 CT and TP53 KO untreated or treated with 5-FU + Iri. at the same concentration at the indicated time points (n = more than 490 cells examined over 3 independent experiments). M Representative γH2A.X staining and quantification of HCT116 and DLD1 cell lines treated for 72 h with 5-FU + Iri. at the same concentration (5-FU 0.1 µg/mL and Iri. 0.04 µg/mL) (n = more than 50 cells examined over 2-3 biologically independent experiments). For all applicable figure panels, data are mean ± SD, except for G, I, J, and L (Tukey method for box plots), where boxes represent the central 50% of the data (from the lower 25th percentile to the upper 75th percentile), lines inside boxes represent the median (50th percentile), and whiskers are extended to the largest value less than the sum of the 75th percentile plus 1.5 IQR (the difference between the 25th and 75th percentile) or greater than the 25th percentile minus 1.5 IQR, and plot any values that are greater or lower than this as individual points. Significance (p) was calculated with one-way ANOVA test, except for B and C (two-way ANOVA) and F (two-sided logistic regression trend test). For G, I, and J scale bar represents 50 μm. ****p < 0.0001; n.s., no significant. 5-FU, 5-fluorouracil; Iri, irinotecan; SA-β-Gal, SA-β-Galactosidase; IC20, IC30, and IC60 indicate 5-FU + Iri. treatment leading to 20, 30, and 60% cell death, respectively. Source data are provided as a Source Data file. in vitro of the general IC20-treated PDO5 population and purified untreated controls (Fig. 2G), as expected. Importantly IC20 and quiescent cells. For this, we generated a PDO5 line (PDO5-hG) IC30-treated PDO cells displayed a significantly higher ability to carrying a doxycycline-inducible histone H2B-GFP reporter that generate intraperitoneal implants when compared with untreated is specifically retained by the quiescent tumor population after tumor cells (Fig. 2G–I). Still, we detected a reduction in the doxycycline withdrawal20. Upon 6 days of doxycycline treatment, proliferation capacity of CT-treated PDO5 cells as determined by PDO5-hG cells were treated with 5-FU + Iri. for 72 h and, after IHC analysis of the proliferation marker ki67 (Fig. 2J, K). Parallel 2 weeks of doxycycline washout, analyzed by flow cytometry and in vivo experiments comparing IC20-treated PDO5, PDO4, and GFPhigh and GFPlow were sorted (Fig. S2B). We found that sorted PDO8 cells indicated a defective capacity of p53 mutant cells to GFPhigh, which represents the quiescent population of CT-treated generate in situ tumors and intraperitoneal implants after cells, displayed an identical capacity for an organoid generation as sublethal CT treatment (Fig. 2L), which was in agreement with GFPhigh plus GFPlow cells (Fig. S2C) indicating that TIC activity is their defective TIC in the in vitro assays. retained in the PQL population. By transplantation of 5000 These results indicate that TP53 WT PDO5 cells show reduced GFPhigh or GFPlow PDO5 sorted cells in the cecum of nude mice, capacity to proliferate in vitro and in the primary tumors after we confirmed the higher clonogenic activity of the quiescent treatment with sublethal 5-FU + Iri. but comparable TIC as population in vivo. In particular, 3 out of 5 mice transplanted untreated cells in vitro and higher metastatic activity in vivo. with GFPhigh cells developed tumors in the cecum and/or intra- peritoneal implants compared with 1 out of 5 mice transplanted CT-induced PQL cells display fetal intestinal stem cells (feISC) with GFPlow cells after 2 months (Fig. S2D). characteristics. Sublethal CT treatment has been linked to the We next studied the in vivo tumorigenic and metastatic acquisition of specific stem cell signatures in B-cell lymphoma10 capacity of IC20 and IC30-pre-treated PDO5 cells using two and intestinal cancer21. To study the transcriptional changes complementary strategies. First, we performed intracardiac associated with the PQL phenotype, we performed RNA injection of 40,000 single cells (untreated, IC20 or IC30 pre- sequencing (RNA-seq) of control, IC20- and IC30-treated PDO5, treated) labeled with firefly luciferase into NOD-SCID-gamma and IC30-treated PDO66 cells. Bioinformatic examination of data (NSG) immunocompromised mice. Mice were analyzed weekly demonstrated a significant overlap between differentially using bioluminescence to monitor metastatic growth using the expressed genes (DEG) in PDO5 and PDO66 (Supplementary IVIS animal imaging system (Fig. 2D). We found that PDO5 Data 1). Moreover, DEG in IC20- and IC30-treated PDO5 showed treated with 5-FU + Iri. displayed a superior and dose-dependent, an almost perfect correlation of gene expression in pairwise although non-significant, metastatic capacity than untreated cells comparisons (IC20 vs. untreated and IC30 vs. untreated) (logistic regression trend test, p = 0.108). Specifically, 7 of 14 mice (p < 2.2e–16, R = 0.974) (Fig. S3A). Principal Component Ana- transplanted with untreated PDO5 cells showed visible metastasis lysis (PCA) indicated that untreated PDO5 and PDO66 clustered 15 weeks after injection compared with 4 of 6 mice transplanted together and formed separate entities when compared with IC20- with IC20-treated cells and 9 of 11 mice with IC30-treated cells and IC30-treated PDO5 or PDO66 cells (Fig. 3A). Gene Set (Fig. 2D, E). Quantitative analysis of the evolution of lesions in an Enrichment Analysis (GSEA) of genes differentially expressed in independent assay demonstrated a significantly higher capacity of CT-treated PDO5 (Fig. 3B) and PDO66 (Fig. 3C) uncovered p53 IC30-treated cells for metastasis initiation (Fig. 2F). as the main activated pathway in CT-treated cells, which was Next, we inoculated equivalent numbers of untreated, IC20 and confirmed in PDO5 by WB analysis (Fig. 3D), qPCR (Fig. 3E) and IC30 pre-treated PDO5 cells in the cecum of athymic nude mice. ChIP assay (Fig. S3B) of canonical p53 targets. DEGs genes are Tumor growth was assessed by palpation weekly and animals also clustered in the NF-κB, EMT, and the interferon-gamma were sacrificed synchronously 70 days after transplantation. We (IFNγ) pathway (Fig. 3B, C), which has been associated with found that untreated, IC20 and IC30-treated PDOs all generated inflammatory response and stemness22–26. Conversely, the E2F tumors at the site of inoculation, being IC20 and IC30-treated pathway that coordinates cell cycle progression at the G1/S derived tumors significantly smaller than those arising from transition (reviewed in ref. 27) and the G2/M checkpoint were 4 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE A B C PDO5 (p53WT) PDO66 (p53WT) PDO4 (p53MUT) PDO8 (p53MUT) PDO5 (p53 KO) n.s. n.s. p=0.0003 **** n.s. p=0.0177 **** **** # of organoids **** # of organoids # of organoids n.s. Diameter (μm) # of organoids Diameter (μm) # of organoids 125 80 **** 10 80 100 n.s. **** 100 60 8 60 80 75 6 60 2 weeks washout 2 weeks washout 40 40 50 4 40 25 20 2 20 20 0 0 0 0 0 p=0.0269 p=0.0002 **** **** Diameter (μm) Diameter (μm) Diameter (μm) 200 **** 150 n.s. * p=0.0052 200 **** 300 **** 300 150 **** 150 250 100 200 100 100 150 50 **** 50 100 50 100 0 0 0 0 0 t. t. 30 30 20 20 t. t. t. 30 30 20 20 30 on on on on on IC IC IC IC IC IC IC IC IC C C C C C D E Control IC20-treated IC30-treated % mice at week 15 100 Time 0 2 80 60 40 > 20 Week2 0 RIP t. 30 20 tre IC IC Un Healthy With metastasis > p=0.10 Week4 > F Rel. photon flux (log10) > 100 > > > > Week8 10 > > 1 * 0.1 Week12 > 104 RIP > > > 0 14 21 28 35 Days after injection 9 Ctrl (n=6) Color Scale IC 30 (n=5) Week15 Min=7.85e3 Max=2.05e4 RIP RIP 8 Radiance (p/sec/cm2/sr) n.s. G p=0.0016 H J In situ tumors K 8 p=0.0339 IP implants In situ Tumors Total tumor weight (gr) p=0.0228 In situ 12 Number of implants 6 Implants p=0.001 Untreated % Ki67+ tumor cells 100 p=0.0043 8 4 80 60 2 4 40 500 μm 500 μm 20 0 0 0 5-FU+Iri. IC20 Control IC20 IC30 . 30 20 at tr. 20 30 re IC IC on nt IC IC U C I Untreated Pretreated IC30 IP implants n.s. 500 μm 500 μm IP implants 0.0205 % Ki67+ tumor cells 100 p=0.002 80 5-FU+Iri. IC30 60 40 20 0 t. 30 20 a 500 μm 500 μm re IC IC nt L U p=0.113 p=0.032 Total tumor weight (gr) 4 3 2 1 0 tr. IC . IC . tr tr C 0 C 0 30 on on on 3 3 IC C PDO5 PDO8 PDO4 among the highest downregulated pathways, together with the are instrumental in the fetal ISC (feISC) after intestinal MYC pathway (Fig. 3B, C), suggesting a general inhibition of injury16,26,29,30, and downregulated such in the case of canonical proliferation. adult ISC markers LGR5 and EPHB2 (Fig. 3G). Accordingly, Unexpectedly, our analysis identified an inversed correlation GSEA indicated a significant correlation between the CT-induced between DEG in PDO5 and PDO66 and the canonical Lgr5+ ISC signature and the transcriptional program associated with fetal signature28 (Fig. 3F). More in-depth analysis showed a mixed ISC conversion and loss of adult Lgr5+ cells30 (Fig. 3H, I), pattern of genes upregulated such as LY6D and YAP1, which which was found to be YAP1-dependent in the regenerating NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 Fig. 2 TP53 WT PQL cells retain tumor-initiating capacity in vitro and in vivo. A–C Number of PDOs (upper panels) and diameter (lower panels) of TP53 WT PDO5 and PDO66 A, TP53 mutant PDO4 and PDO8 B, and TP53 KO PDO5 C treated with 5-FU + Iri. as indicated, and left for 2 weeks with fresh medium. 300 cells/well were seeded (n = 6 wells examined for TICs and n = more than 50 spheres examined whenever possible for diameters, from three biologically independent experiments). D Bioluminescence images of NGS mice after intracardiac injection of 40,000 luciferase-PDO5 CT and IC20 or IC30- treated cells (from 1 out of 2 biologically independent experiments). E Percentage of healthy and metastasis-carrying mice at week 15 (n = 14 (Untreat.), 6 (IC20), and 11 (IC30) mice examined over two biologically independent experiments). F Relative photon flux measurement of metastasis initiation in mice injected with PDO5 CT and IC30-treated cells (n = 6 (Untreat.) and 5 (IC30) mice examined from one experiment). G–I Total tumor weight of in situ tumors and intraperitoneal implants per animal in the different experimental groups G (n = 5 mice examined over two biologically independent experiments), number of intraperitoneal implants H (n = 3 mice examined from 1 experiment), and photographs of tumors derived from orthotopically implanted CT, IC20 and IC30-pre-treated PDOs in nude mice I. J, K IHC analysis of ki67 in in situ tumors and implants J and percentage of ki67+ cells in the indicated conditions K (n = more than four independent regions examined). L Tumor weight of intraperitoneal implants of tumors derived from orthotopically implanted CT and IC30-pre-treated PDO5, PDO8 and PDO4 in nude mice (n = 2 (Untreat.) and 5 (IC30) mice examined from 1 experiment). For all applicable figure panels, data are mean ± SD. Significance (p) was calculated with one-way ANOVA test, except for C (two-sided Student’s T test), E (two- sided logistic regression trend test) and F and G (two-way ANOVA test). ****p < 0.0001; n.s., no significant. 5-FU, 5-fluorouracil; Iri, irinotecan. Source data are provided as a Source Data file. intestine29,31. We used the PDO5-hG line to test whether obtained at diagnosis with available clinical data (Supplementary upregulation of genes contributing to feISC and EMT pathways Table S6). We determined the H-score of nuclear YAP1 as upon sublethal CT was present in the quiescent cell population. intensity multiplied by the percent of positive tumor cells in the Cells were treated as explained before (see Fig. S2B), sorted based triplicates, and stratified patients accordingly. Considering the on GFP levels, and processed for qPCR analysis. We detected a mean value ± 0.2 standard deviations of the H-score, we observed massive upregulation of feISC and EMT genes in CT-treated cells a trend towards poor prognosis in the group with higher nuclear that were restricted to the GFPhigh population (Fig. 3J). YAP1 (disease-free survival: p = 0.26; HR = 1.38) that increased when considering mean value ± 0.4 s.d. (p = 0.12; HR = 1.58). Acquisition of feISC by CT treatment is linked to and Prognosis value of nuclear YAP1 levels reached statistical dependent on YAP1 activation. We investigated whether feISC significance when considering the mean value ± 0.6 standard conversion of human CRC cells induced by CT was imposed by deviations (p = 0.039; HR = 1.97) (Fig. 4H, I). Indicative of the YAP1 signaling. By WB analysis of PDO5 cells, we detected clinical applicability of our observations, verteporfin treatment increased YAP1 expression after 72 h of 5-FU + Iri. treatment increased the sensitivity of PDO5 cells to 5-FU + Iri. (Fig. 4J). (Fig. 4A). In addition, we detected an accumulation of nuclear Further suggesting that verteporfin effects are linked to YAP1 (active) YAP1 in IC20 and IC30-derived PDO5 tumors 2 months inhibition, genetic YAP1 deletion precluded PDO5 clonogenicity after implantation in mice (Fig. 4B). We studied whether YAP1 in TIC assays (Fig. 4K) associated with increased basal and CT- activity was required for transcriptional induction of feISC genes induced DNA damage (Fig. 4L). in PDO cells after CT. Incubation of PDO5 cells with the YAP1 Together these results indicate that feISC conversion imposed inhibitor verteporfin precluded induction in several randomly by CT is YAP1 dependent. Moreover, detection of nuclear YAP1 selected fetal genes following IC20 5-FU + Iri. treatment (Fig. 4C). in CRC tumors is predictive of poor prognosis, which could be Indicating the specificity of verteporfin effects, a PDO5 YAP1 KO therapeutically exploited. pool generated by CRISPR-Cas9 (Fig. S4A) showed comparable impairment of feISC genes induction by CT (Fig. 4D). A restricted YAP1-dependent fetal signature shows coordinate To study whether a similar mechanism operates in patients, we expression in human CRC associated with higher p21 levels. took advantage of a set of paired CRC samples (n = 62) collected The presence of nuclear YAP1 in untreated tumors associated at diagnosis (biopsy) and at the time of surgery (n = 62) from with poor prognosis, led us to speculate that a YAP1-dependent tumors showing partial response after DNA-damaging-based fetal signature could already be present in tumors at diagnosis. neoadjuvant treatment (Supplementary Table S2). By IHC We took advantage of public data sets to interrogate the inci- analysis, we detected nuclear YAP1 in the epithelial component dence and prognosis value of feISC signature in untreated CRC of 10 out of 46 tumors at diagnosis, considering positive those human tumors. Computational analysis of the Marisa32 carrying ≥20% positive cells. Notably, the number of tumors (GSE39582), Jorissen33 (GSE14333) and TCGA (TCGA Portal) carrying nuclear YAP1 was massively increased after neoadjuvant CRC data sets using CANCERTOOL34 indicated differential treatment (35 out of 45 analyzed) (Fig. 4E and Supplementary expression of most feISC genes with some of them distributed Table S2), which was associated with the expression of the feISC in clusters of coordinated expression (with either positive or markers S100A4 and SERPINH1 (Fig. 4F). Neoadjuvant treat- negative correlation) (Supplementary Data 2, in the Marisa data ment was also linked to a general reduction in tumor cell set). We integrated feISC genes with the highest coordinated proliferation, as determined by IHC analysis of ki67, in the expression in a new cluster containing 28 plus 8 genes that were absence of senescence traits such as enlarged nuclei (Supplemen- either upregulated (28up) or downregulated (8down) in CT- tary Table S2) or high p16 levels (Fig. S4B). These characteristics treated PDO cells and fetal ISCs (Fig. S5A). The 28up + 8down- were suggestive of human CRC tumors experiencing PQL feISC gene signature was present in Marisa (Fig. 5A), Jorissen, conversion after sublethal CT. However, the limited number of and TCGA CRC cohorts (Fig. S5B) associated with higher levels samples in this cohort and the fact that a majority of post- of the cell cycle regulator p21 (CDKN1A gene) (Fig. S5C), treatment samples displayed nuclear YAP1 accumulation and further suggesting a link between fetal traits and slow-cycling proliferation inhibition precluded establishing the potential phenotype. We confirmed upregulation of several genes of the prognosis value of these parameters. 28up + 8down-feISC signature by RT-qPCR (Fig. 5B) and WB Since nuclear YAP1 was also detected in untreated tumors, we analysis (Fig. 5C) of IC20 5-FU + Iri.-treated PDO5. Activation performed IHC analysis of YAP1 (Fig. 4G) in a tissue microarray of fetal genes following CT treatment was comparably observed containing 196 different human CRC samples (in triplicates) in PDO66 (Fig. 5D) and slightly reduced in the hypomorphic 6 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE A B PDO5 P53_PATHWAY PDO5 CT.1 10 PDO66 IC30.1 PDO5 CT.2 TNFA_SIGNALING_VIA_NFKB PC2: 1% variance PDO66 IC30.2 MYOGENESIS 5 EPITHEL_MESENCH_TRANSITION KRAS_SIGNALING_UP 0 PDO66 CT.1 PDO66 IC30.3 PDO5 IC20.1 PDO66 CT.2 INTERFERON_GAMMA_RESPONSE PDO5 IC20.2 PDO66 CT.3 MYC_TARGETS_V1 −5 PDO66 IC30.4 PDO66 CT.4 PDO5 IC30.1 E2F_TARGETS −10 PDO5 IC30.2 G2M_CHECKPOINT −20 0 20 40 PC1: 99% variance -3.0 -2.0 -1.0 0 1.0 2.0 Normalized Enrichment Score C D ol PDO66 ntr P53_PATHWAY IC20 treatment (h) Co TNFA_SIGNALING_VIA_NFKB 1 3 6 24 48 72 IC60 (kDa) EPITHEL_MESENCH_TRANSITION WB: INPLAMMATORY_RESPONSE p53 50 KRAS_SIGNALING_UP INTERFERON_GAMMA_RESPONSE p21 20 MYOGENESIS MYC_TARGETS_V1 G2M_CHECKPOINT Tubulin 50 E2F_TARGETS -3.0 -2.0 -1.0 0 1.0 2.0 Normalized Enrichment Score E 10.0 G Control IC20 CDCA7 Relative mRNA levels *** MEX3A 7.5 * LGR5 **** ** *** EPHB2 *** CD44 5.0 AXIN2 ASCL2 **** HES1 1 2.5 LY6D 0 YAP1 EPHA4 −1 SOX9 0.0 −2 PROM1 1 PH 2 N1 3 A3 3 LA 1 K2 5 DM p2 SN M DK P5 AT . . 30 30 30 20 LD PL nt nt PT IC IC C IC A/ T ZM SE M U U I 66 66 5 5 66 5 O O O O O PD PD O PD PD PD PD C F Enrichment Plot: H I Enrichment Plot: Enrichment Plot: Intestinal Stem (Muñoz et al. ) Fetal DOWN (Mustata et al.) Fetal UP (Mustata et al.) Enrichment Score (ES) Enrichment Score (ES) Enrichmen Score (ES) NES: -1.85 NES: -1.53 0.40 NES: 1.42 0.00 0.10 p-value: 0.0000 p-value: 0.000 0.30 p-value: 0.003 -0.10 0.00 0.20 -0.20 -0.10 0.10 -0.20 -0.30 0.00 -0.30 -0.05 -0.40 T positive correlated C negative correlated T positive correlated C negative correlated T positive correlated C negative correlated J Fetal ISC genes EMT genes **** 15 *** 150 Untreated n.s. **** IC20 GFP+ Relative mRNA levels **** Relative mRNA levels **** **** 125 IC20 GFP- **** **** 10 n.s. 100 **** **** **** 75 **** 5 50 25 0 0 2 1 E1 AI 1 1 BS 1 6 AM 8A PC BB N SN TH PI L1 G TU IC R O SE C Fig. 3 PQL phenotype is associated with the acquisition of a fetal intestinal stem cell (feISC) signature. A PCA of the RNA-seq analysis from the indicated replicates performed on PDO5 and PDO66 (untreated; CT) or 5-FU + Iri.-treated either at IC20 or IC30 as indicated. B, C Barplots depicting the normalized enrichment score of statistically significant enriched pathways obtained by GSEA analysis of PDO5 (B) and PDO66 (C) with the Hallmark gene set for treated samples (NOM p val < 0.05). D WB analysis of control and 5-FU + Iri.-treated PDO5 cells collected at the indicated time points (from one out of three biologically independent experiments). E RT-qPCR analysis of selected p53 target genes from control and IC20-treated PDO5 cells (n = 4 biologically independent experiments). F GSEA of an intestinal stem cell (ISC) gene set, according to Muñoz et al., in untreated (C) versus treated (T) PDO5/PDO66 condition. G Heat map showing the expression levels of the indicated ISC genes in untreated, IC20 and IC30-treated PDO5 cells and untreated and IC30-treated PDO66 cells. H, I GSEA of a fetal down (H) and a fetal up (I) stem cell gene set, according to Mustata et al., in control (C) versus treated (T) PDO5/PDO66 condition. J RT-qPCR analysis of selected fetal ISC and EMT genes from untreated and IC20-treated sorted GFPhigh or GFPlow PDO5 cells (from 1 out of 2 biologically independent experiments). For all applicable figure panels, data are mean ± SD. Significance (p) was calculated in E with two-sided Student’s t test and in J with one-way ANOVA test. For F, H, I p.value is nominal p value given by the GSEA program. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; n.s., no significant; CT, control; IC20 and IC30, 5-FU + Iri. are doses imposing 20 and 30% cell death, respectively; GSEA, gene set enrichment analysis; NES, normalized enriched score. Source data are provided as a Source Data file. TP53 mutant PDO4 (Fig. 5E) and the TP53 KO PDO5 pool carrying WT or mutant TP53 (Fig. 5H) but the massive accu- (Fig. 5F). At the protein level, we detected slight differences in mulation of apoptotic cells as determined by cCas3 and the activation of several feISC markers (such as TIMP2, MRAS, cPARP1 levels (Fig. 5G, S1H, and S1I). By ChIP-seq assay of 5- ICAM, or TSPAN4) after 5-FU + Iri. treatment when com- FU + Iri. IC20-treated PDO5 cells, we only detected 3 genes in paring TP53 WT and KO PDO5 (Fig. 5G) or CRC cell lines the 28up-feISC signature, PLK2, PHLDA3 and GSN, that were NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 A B Control IC20 IC30 > > YAP1/DAPI > 5-FU+iri PDO5 > (hours) 0 72 > >> > (kDa) > > > WB: > YAP1 75 >> 100 **** > 50 μm 50 μm 50 μm % nucler YAP1 pos. cells TP53 50 75 > > YAP1/Ki67/DAPI Histone H3 15 > 50 > p=0.0138 > >> > > > > 25 > >> > 50 μm 50 μm 50 μm 20 30 r. nt IC IC U C CT D n.s. WT untreated 8 8 WT IC20 n.s. IC20 KO untreated Relative mRNA levels IC20 n.s. Relative mRNA levels * KO IC20 + Verteporfin 6 ** 6 n.s. * * * n.s. n.s. * n.s. * ** * ** ** n.s. n.s. 4 n.s. ** * 4 ** * * * * ** ** 2 2 0 0 L2 PH 1 AS 1 6 2 1 1 A3 4 1 4 1 1 TS 2 AS 1 L2 6 8A BB M AM A3 H PC N PC N 8A H M AM 99 BB LD N PA R TP PA 99 L1 N R TU TP G LD PI M IC D L1 G TU PI TS M IC O C R D PH O C SE R C C SE E F YAP1/DAPI SERPINH1/DAPI IHC: YAP1 Patient #24 Biopsy 50 μm 50 μm YAP1/DAPI S100A4/DAPI Post-neoadj. Patient #58 50 μm 50 μm 50 μm Patient #3 Patient #14 Patient #42 50 μm 50 μm G H I IHC: YAP1 1.00 1.00 Disease−free survival Overall Survival 0.75 0.75 0.50 0.50 0.25 nYAP1 < mean -0.6 SD 0.25 nYAP1 < mean -0.6 SD nYAP1 > mean+0.6 SD nYAP1 > mean+0.6 SD HR=1.97 HR= 2.26 50 μm p = 0.039 p = 0.040 Time 0 Time 0 (months) 0 25 50 75 100 (months) 0 25 50 75 100 J K **** L # of organoids 60 120 PDO5 Relative Cell Growthl IC20 > 5-FU+Iri. IC20 > Verteporfin 40 p=0.0136 100 IC20 > 5-FU+Iri. 120 Tail moment 80 +Verteporfin 20 100 0 80 60 60 **** 40 40 Diameter (μm) 200 20 20 150 0 72 h. h. 3 . U . r r nt nt 0 100 U -3 -2 -1 0 1 2 3 50 YAP1 KO 0 Log concentration (mg/ml) t. 30 on IC C Fig. 4 Acquisition of feISC by CT treatment is dependent on YAP1 activation. A WB analysis of PDO5 cells was collected after 72 h of 5-FU + Iri. treatment (from one out of three biologically independent experiments). B Representative images and quantification of YAP1 and ki67 staining in tumors derived from control (CT), IC20, and IC30-pre-treated PDOs implanted in nude mice. White arrows indicate cells with nuclear YAP1. DAPI was used as a nuclear marker (n = 9 independent regions examined). C RT-qPCR analysis of randomly selected feISC genes in control and 5-FU + Iri-treated PDO5 alone or plus 0.2 μM verteporfin (n = 2 biologically independent experiments). D RT-qPCR analysis of same genes in control and YAP1 KO PDO5 treated as in C (n = 2 biologically independent experiments). E, F Representative images of YAP1 E and SERPINH1 and S100A4 F in the indicated tumors at diagnosis or after neoadjuvant treatment (from 1 out of 6 biologically independent samples). G Representative image of an untreated human CRC identified as nuclear YAP1 high (done with all TMA samples, described in Supplementary Table S6). H, I Kaplan–Meier representation of disease-free H and overall I survival of CRC patients classified according to the H-score of nuclear YAP1. J Dose–response curves of IC20-pre-treated PDO5 and then treated for 3 days as indicated (n = 3 replicates examined, from one out of three biologically independent experiments). K Number of PDOs (upper panel) and diameter (lower panel) of PDO5 YAP1 KO after 72 hours of 5-FU + Iri. treatment and 2 weeks of washout. 300 cells/well were seeded (n = 6 wells examined for TICs and n = more than 50 spheres examined whenever possible for diameters, from three biologically independent experiments). L Comet assay in PDO5 YAP1 KO treated with 5-FU + iri. for 3 and 72 hours as indicated (n = >790 cells examined over three biologically independent experiments). For all applicable figure panels, data are mean ± SD, except for L (Tukey method for box plots), where boxes represent the central 50% of the data (from the lower 25th percentile to the upper 75th percentile), lines inside boxes represent the median (50th percentile), and whiskers are extended to the largest value less than the sum of the 75th percentile plus 1.5 IQR (the difference between the 25th and 75th percentile) or greater than the 25th percentile minus 1.5 IQR, and plot any values that are greater or lower than this as individual points. Significance (p) was calculated with one-way ANOVA, except for D and K by two-sided Student’s t test and for H and I by log-rank (Mantel–Cox) test. *p < 0.05, **p < 0.01, n.s., no significant. CT, control; IC20 and IC30, 5-FU + Iri. treatment leading to 20 and 30% cell death, respectively. Source data are provided as a Source Data file. 8 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE A B C 5-FU+iri SERPINH1 COL18A1 SLC27A2 LAPTM5 PHLDA3 CXCL16 TSPAN4 (hours) 0 1 2 6 24 48 72 (kDa) CD99L2 S100A4 GLIPR1 HOOK1 AGMAT KCNK5 ABHD4 VAMP5 PLAUR PDSS1 ANXA1 ARL4C TUBB6 ICAM1 TIMP2 CRIP2 RHOD MRAS HUNK KIFC3 IL1RN GPC1 CDX1 TPM2 WTIP PLK2 Untreated MYB GSN CLU 6 *** WB: 100 1 GLIPR1 IC20 ICAM1 Relative mRNA levels ANXA1 PLK2 0.8 PHLDA3 S100A4 CXCL16 TIMP2 25 oL IL1RN 4 ** ** ** 0.6 PLAUR *** ** ** 37 KIFC3 WTIP ** MRAS 0.4 GPC1 RHOD ** ** ABHD4 50 0.2 GSN SERPINH1 2 SERPINH1 COL18A1 CRIP2 0 TIMP2 TUBB6 * * MRAS TSPAN4 * TUBB6 50 −0.2 ARL4C 0 LAPTM5 ICAM1 CD99L2 X1 TU 9L2 M 1 PI S LI 2 SE M C3 D 1 TP B6 PA 1 AR M1 VA L4C L1 1 KC AT IC N4 C GP 5 C 5 C P2 KI P2 AG 8 A G M C NH TS P R O C K P VAMP5 R RA −0.4 D B N M M I F 37 A CD99L2 R 9 TI CLU TPM2 −0.6 HUNK KCNK5 p21 20 SLC27A2 HOOK1 −0.8 MYB Tubulin 50 CDX1 AGMAT −1 PDSS1 D 10.0 ** PDO66 (P53 WT) E 4 Untreated PDO4 F 8 **** * PDO5 n.s. Untreat. IC20 Untreated ** IC20 (P53 hyp. I254T) n.s. KO#3 Untreat. Relative mRNA levels Relative mRNA levels Relative mRNA levels IC20 ** * KO#3 IC20 7.5 3 n.s. 6 ** ** n.s. **** n.s. *** n.s. 5.0 ** * 2 * 4 *** ** * * ** ** n.s. n.s. n.s. 2.5 * n.s. n.s. 1 * 2 * 0.0 0 0 PH 8 A 1 T 6 C H1 TU L2 G 1 A3 L1 1 PA 2 PI S IC N4 S E M 1A PH A 1 T 6 C H1 TU L2 G 1 A3 L1 1 PA 2 PI S IC 4 S E M 1A BB AM C PC TS PM R RA TP 1 PH 1A BB AM 1 C 8A1 C PC TS PM N TU 3 R AS R RA L1 1 2 IC 6 99 4 LD TS 9 L 2 AM N KN 99 H 8 LD A N KN C M BB N KN LD N SE MR D P D PA D O D PI G 9 O C D C D O C C G H I PDO5 P53 WT P53 MUT 5-FU+Iri. p53 WT p53 KO HCT116 LS174T HT29 SW480 5% 2% Subtype (hours) 0 3 6 24 48 72 0 3 6 24 48 72 (kDa) 5FU+Iri. IC30 - + - + - + - + (kDa) CMS1 WB: WB: 21% 25 34% CMS2 TIMP2 ICAM1 75 64% CMS3 74% CMS4 MRAS SERPINH1 50 NOLBL 25 ICAM1 TSPAN4 25 28up-high/8down-low 28up-low/8down-high 75 MRAS 37 Guinney et al., 2015 YAP1 75 S100A4 20 10 cCasp3 p21 20 cPARP1 75 Tubulin 50 Tubulin 50 p21 20 J Lee et al., 2020 Epithelial subtypes S100A4 PLAUR ANXA1 5 5 5 5 9 8 tSNE_2 tSNE_2 tSNE_2 tSNE_2 3 2 0 4 7 0 0 0 6 1 5 -5 -5 -5 -5 -10 -10 -10 -10 -5 0 5 10 -5 0 5 10 -5 0 5 10 -5 0 5 10 tSNE_1 tSNE_1 tSNE_1 tSNE_1 ARL4C GSN CXCL16 TIMP2 5 5 5 5 tSNE_2 tSNE_2 tSNE_2 tSNE_2 0 0 0 0 -5 -5 -5 -5 -10 -10 -10 -10 -5 0 5 10 -5 0 5 10 -5 0 5 10 -5 0 5 10 tSNE_1 tSNE_1 tSNE_1 tSNE_1 direct p53 targets (Fig. S5D). These results suggested that p53 and collaborators35. 74% of tumors with the 28up + 8down-feISC does not participate in the transcriptional activation of feISC signature were categorized as CMS4 (Fig. 5I), which is genes but might favor survival of cancer cells experiencing fetal characterized by upregulation of epithelial-to-mesenchymal conversion. transition (EMT) gene signatures, TGFβ signaling, stromal We also determined whether tumors carrying the 28up + 8 infiltration, and poorer patient prognosis. In contrast, tumors down-feISC signature were restricted to a specific cancer displaying the opposite phenotype (28up-low+8down-high) were molecular subtype (CMS), based on the classification by Guinney primarily ascribed to the more canonical Wnt and Myc-driven NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 Fig. 5 CT-induced quiescent cells display a fetal intestinal stem cell signature that is partially dependent on p53. A Expression correlation matrix from the 28up + 8down-feISC gene signature in Marisa database (n = 566). Size of circles and color intensity are proportional to Pearson correlation coefficient for each gene pair. B RT-qPCR analysis of normalized expression of selected 28up + 8down-feISC genes in untreated and treated PDO5 (n = 2–4 biologically independent experiments). C WB analysis of control and 5-FU + Iri-treated PDO5 cells collected at the indicated time points (from one out of three biologically independent experiments). D–F RT-qPCR analysis of normalized relative expression of indicated genes in control and CT-treated TP53 WT PDO66 D, TP53 mutant PDO4 E and PDO5 TP53 KO #3 F (n = 2–4 biologically independent experiments). G WB analysis of PDO5 and PDO5 TP53 KO untreated or treated with 5-FU + Iri. at the same concentration at the indicated time points (from one out of three biologically independent experiments). H WB analysis with indicated antibodies of CRC cells untreated or treated 72 h with 5-FU + Iri. (from one out of three biologically independent experiments). I Pie charts showing the molecular subtype distribution according to Guinney et al., in patients within the feISC signature groups as indicated. J Distribution of selected 28up-fetal-ISC genes in epithelial subtypes cell states 1–9 described by Lee et al.36. The t-SNE plots were obtained using the web-based tool URECA (User-friendly InteRface tool to Explore Cell Atlas). For all applicable figure panels, data are mean ± SD. Significance (p) was calculated with a two-sided Student’s t test. *p < 0.05; **p < 0.01; ****p < 0.0001; n.s., no significant. CT, control; IC20 and IC30, 5-FU + Iri. treatment indicates 20 and 30% cell death, respectively. Source data are provided as a Source Data file. CMS2 subtype. We studied whether the feISC signature of Together with our results demonstrate the existence of a untreated tumors was expressed in the epithelial cancer cells or YAP1-dependent feISC signature that is induced after sublethal primarily contributed by the stromal component. Analysis of CT and favors cancer progression and metastasis. This single-cell RNA-seq data from Lee and collaborators36 demon- signature predicts poor survival of CRC patients in the context strated that feISC genes are expressed in the epithelial cancer of functional p53. cells, particularly in states 1, 5, and 6 from Lee and collaborators that are all associated with the secretory and migratory pathways Discussion (Fig. 5J). We have here identified a YAP1-dependent feISC signature that These results identified a specific feISC signature that is can be induced by CT, associated with the acquisition of a PQL induced by sublethal CT in a YAP1-dependent manner, but it is state, but that is already present in untreated tumors from several already present in the subset of CMS4 subtype (untreated) CRC CRC cohorts. The molecular mechanisms inducing YAP1 acti- tumors from Guinney and collaborators, and in the secretory and vation by CT are not yet known, and we speculate that stromal migratory epithelial states 1, 5, and 6 from Lee and collaborators. populations present in the tumor, such as inflammatory cells, may induce upstream regulators of this signature (i.e., TGFβ signaling) thus leading to the acquisition of PQL traits before CT The YAP-dependent feISC signature is predictive of reduced treatment. In agreement with this possibility, tumors carrying the disease-free survival in TP53 WT tumors. Our results indicated 28up + 8down-feISC signature are primarily included in the that feISC conversion was induced by YAP signaling, and nuclear CMS4 CRC subtype from Guinney and collaborators35 char- YAP1 predicts a poor prognosis in untreated human CRC. Thus, acterized by stromal infiltration and TGFβ signaling. It was we studied whether the 28up + 8down-feISC signature can be shown that TGFβ can promote YAP1 signaling by facilitating the used as a prognosis tool in patients. The global 28up + 8down- degradation of the negative regulator of the pathway RASSF1A37. feISC signature was sufficient to demarcate at least two subsets of Moreover, identification of this feISC signature will allow the patients in the Marisa (Fig. 6A and Supplementary Table S3), classification of patients with a higher probability of recurrence at Jorissen, and TCGA data sets (Supplementary Table S3), within diagnosis, which will benefit from more aggressive treatments or the group with the highest 28up and lowest 8down-feISC levels closer follow-up. Additionally, targeting the upstream signals displaying the poorest disease-free-survival (Fig. 6B). A more imposing PQL/feISC acquisition pharmacologically (i.e. YAP1 or detailed analysis of the Marisa data set demonstrated that this TGFβ inhibitors) or using combination treatments that effectively signature was significantly associated with tumor relapse in eradicate quiescent tumor cell populations (i.e. CT plus inhibitors patients at stages II (n = 264) (p = 0.041) (Fig. 6C) and II + III of the NHEJ repair pathway) appear as interesting therapeutic (n = 469) (p = 0.0033) (Fig. 6D), and imposed a trend towards options. Conversion of adult into fetal ISC had already been poor prognosis at stage IV (n = 60) (Fig. 6E). identified as part of the process of tissue regeneration after hel- The presence of functional p53 defines DNA repair efficiency minths infection26 or in the Dextran Sulfate Sodium colitis and TIC of cells acquiring the feISC signature after damage (see model29. Thus, our results reinforce the concept that tumor Figs. 2A, B, 5F, G). Thus, we explored the possibility that TP53 development is partially mimicking the tissue regeneration status determined the prognosis value of 28up + 8down-feISC process. signature in CRC patients. We observed increasing proportion of CT is the current treatment for advanced and metastatic TP53 mutated tumors according to tumor stage in the TCGA data colorectal tumors. However, in a percent of cases, tumor cells set (Fig. S6A) as expected, however stratification of Marisa that escape from treatment (by efficient drug clearance, effec- (Fig. 6F) and TCGA (Fig. S6B) patients according to TP53 status tive DNA repair, or reduced accessibility of the drugs) can did not have a prognosis value by itself. Importantly, TP53 status acquire a dormant phenotype that could provide superior reliably determines the prognosis value of 28up + 8down-feISC resistance to subsequent damaging-based treatments. In the signature specific at stages II + III (Fig. 6G and Fig. S6C). Because present study we have shown that sublethal CT imposes a non- 28up + 8down-feISC tumors are mainly included in the worst senescent and non-proliferating phenotype on cancer cells, in prognosis CMS4, we tested whether this feISC signature the absence of persistent DNA damage (we refer to this phe- represents an independent prognosis factor inside this molecular notype as PQL). subtype. Carrying the 28up + 8down-feISC signature increased Importantly, PQL cells can efficiently escape from dormancy the risk of relapse in patients within the CMS4 group (Fig. 6H). following in vivo transplantation, as it is shown in the intra-cecal Interestingly, we observed a slight accumulation of TP53 WT in xenograft experiments, in particular when cells migrate from the tumors carrying the 28up + 8down-feISC signature in the site of implantation. This is in agreement with their higher subgroups of CMS4 from TCGA and Marisa cohorts (Fig. S6D). metastatic potential and is also in accordance with previous 10 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE A B 1.00 Marisa, all stages Disease−free survival + + ++ + + + + ++ + + ++ ++++ + + + + +++ 0.75 + ++ + + +++ ++ + + + ++ ++ + ++ ++ ++ ++++ +++ ++ +++++++ ++++ ++ +++ ++++ ++ +++ ++ +++ ++ + + + + + + + + ++ +++ ++ ++ ++ ++ ++ ++ ++ +++ + Stage + +++ ++ ++ +++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++++ ++ ++ + +++ + + ++ +++ ++ + + ++ + + + + + + + + + + + 0.50 + +++++++ ++ +++ ++ +++ +++ ++ + + + 0.25 28up=high/8down=low HR=2.33 28up=low/8down=high p=8e-04 Unclassified Time 0 (months) 0 50 100 150 200 1.00 Jorissen Disease−free survival + +++ ++ + ++ ++++ +++ + + ++ + + +++ ++ + + +++ + ++ + +++ + ++++ + +++ +++ ++ +++ + +++ + ++++ +++ + +++ + + ++ + +++ ++ + + + + ++ + + + + + + 0.75 + +++ + + ++ + + + + + ++ + + +++ + +++++ + + +++ + ++ + + + +++ + +++ + + + 0.50 ++ + + + + 0.25 HR=3.28 28up=high/8down=low p=0.0001 28up=low/8down=high Time 0 Stage (months) 0 50 100 150 IV 1.00 TCGA Disease−free survival III + + + + + + +++ + +++ ++ ++ + + ++++ ++ + ++ + + ++ + + + ++ + + + + + + ++ + II + ++ + + + + ++ + + + + ++ + ++ + ++++ + + + + + + + 0.75 + + + +++ ++ + + ++ + + + + ++ ++++++ + +++ + ++ + + I + + ++ + + +++ + + ++ + + + + + + ++ + ++ + ++ + + + + ++ ++ + ++++ 0.50 + + + + + + ++ + ++ + G AN PR PL XA1 S1 2 PH 0A G LDA R 1 KI OD W C3 TP IP C 2 C SE 9 9 L C PI C L18 H1 TS IP2 1 M AN TU AS TI B6 AR P2 LA L4C IC TM VA M1 G P AB N C D4 PL CL1 IL R SL RN H 27 M OK 2 PD AGSS1 C MA H X1 KC NK + H LU D O N R A X O A D T U LI PC 3 S 5 R 4 1 YB 1 M ++ ++ F T A 5 K 0 P AU 6 C M P R 2 B M H N + + K5 1 4 0.25 + + + HR=2.20 28up=high/8down=low Expression levels p=0.024 28up=low/8down=high -5 0 +5 Time 0 Unclassified (months) 0 50 100 150 C D E 1.00 Marisa, Stage II 1.00 Marisa, Stage II-III 1.00 Marisa, Stage IV Disease−free survival Disease−free survival Disease−free survival + + + + ++ + + + + + +++ ++ + + + +++ + + ++ + + +++ ++++ + + ++ ++ + + ++ + 0.75 0.75 0.75 +++ ++ ++ ++ +++ ++ ++ ++ ++ ++ +++ +++ ++ ++ ++ +++ +++ + ++ + ++ + + + + + + + +++ +++ ++ +++ ++ ++ ++ ++ +++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ ++ +++ ++ ++ ++ ++ ++ +++ +++ ++ + +++ +++++ +++ + + + + + ++ + + + + + + + + + + + ++ + + + + + + + + HR=1.40 p = 0.13 ++++ +++++ +++ ++ + +++++++++ +++ + ++ + 0.50 + + + + 0.50 ++ ++ + + 0.50 + + + + + +++ + + + ++ + + + + 0.25 0.25 0.25 + ++ HR=2.27 28up=high/8down=low HR=1.70 28up=high/8down=low 28up=high/8down=low + p = 0.041 28up=low/8down=high p = 0.0033 28up=low/8down=high 28up=low/8down=high Time 0 Time 0 Time 0 (months) 0 50 100 150 200 (months) 0 50 100 150 200 (months) 0 25 50 75 100 F G 1.00 Marisa 1.00 Marisa (P53 WT) 1.00 Marisa (P53 MUT) Disease−free survival Disease−free survival Disease−free survival + ++ ++ + + + + ++ +++ + + ++ +++ ++ + +++ 0.75 + +++ ++++ + + ++ ++ ++ + 0.75 + + + ++ ++ ++ ++ ++ ++++ +++ + + ++ + + + + + 0.75 ++ ++ +++ + ++ ++ +++ + ++ + +++ +++ ++ ++ ++ ++ ++ ++++++ ++ ++ + + +++ + + + + + + +++++ ++ + ++++ + +++ + +++ + ++ ++ +++ ++ + ++ ++ ++ ++ ++++ ++ + + +++ ++ ++ ++ + + ++ 0.50 0.50 + + ++ + + + + + 0.50 + + + + ++ ++ ++ 0.25 0.25 0.25 p53 WT HR=2.16 28up=high/8down=low 28up=high/8down=low p = 0.08 p53 MUT p = 0.041 28up=low/8down=high p = 0.95 28up=low/8down=high Time 0 Time 0 Time 0 (months) 0 25 50 75 100 (months) 0 25 50 75 100 (months) 0 25 50 75 100 H 1.00 Marisa, in CMS4 Disease−free survival + + 0.75 + + ++ + + ++ ++ ++ +++ ++ +++ +++ ++ +++ ++ ++ + + + + 0.50 + + ++ + 0.25 HR=1.79 28up=high/8down=low p = 0.076 28up=low/8down=high Time 0 (months) 0 50 100 150 200 studies showing that dormant populations of primary human These divergent results were likely indicating that intracardiac CRC cells retain tumor propagation potential38 and cancer cells injection does not reflect all steps of the metastatic process such displaying reversible-quiescent states after CT exhibit increased as EMT or invasive capacity. Supporting this possibility, EMT is tumorigenic potential12. Of note, whereas differences in tumor one of the pathways that are enriched in the CT-treated PDOs dissemination between untreated and CT-treated PDO5 in the (see Fig. 3A), and we have now confirmed that EMT genes are intra-cecal transplantation experiments were highly significant, specifically upregulated in the quiescent cell population (see we only detected slight differences in the intracardiac assays. Fig. 3H). NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 Fig. 6 Identification of a fetal ISC signature with prognosis value in CRC. A Unsupervised hierarchical cluster analysis of patients according to 28up + 8down-feISC gene signature leading to the classification of patients into four subsets (colored in red, green, light blue, and purple). Tumor staging is indicated. B Kaplan–Meier representation of disease-free survival (DFS) over time for patients according to 28up + 8down-feISC signature selected according to A for Marisa (28up = high/8down = low n = 66, 28up = low/8down = high n = 114 and unclassified n = 386), Jorissen (28up = high/ 8down = low n = 114 and 28up = low/8down = high n = 112) and TCGA (28up = high/8down = low n = 39, 28up = low/8down = high n = 96 and unclassified n = 194) colorectal cancer databases. Patients were selected according to cluster analysis of the 28up + 8down-feISC signature. C–E Kaplan–Meier curves representing patients DFS classified according to cluster analysis of the 28up + 8down-feISC signature of stage II (n = 149) C, stage II and III (n = 468) D, and stage IV (n = 60) E patients from Marisa database. F Kaplan–Meier representation of patients DFS classified according to TP53 in the Marisa data set (TP53 WT n = 161 and TP53 MUT n = 190). G Kaplan–Meier representation of DFS of patients from the Marisa data set classified according to cluster analysis of the 28up + 8down-feISC signature in TP53 WT (n = 85) and TP53 mutant (n = 20) groups. H Kaplan–Meier representation of DFS in CMS435 patients from Marisa cohort classified based on 28up + 8down-feISC signature (n = 91). Data in A show normalized, centered, and scaled Illumina probe set intensities on a log2 scale. The stage lane represents the subtype corresponding to each patient. We used Cox proportional hazards models for statistical Kaplan–Meier analysis and log-rank two-sided p value (see Supplementary Table S3). HR, hazard ratio. Our transcriptomic studies revealed that acquisition of PQL linked to TP53 status. Further studies are required to clarify these phenotype rely on p53 and p21 signaling. It has long been discrepancies. Independently on the mechanisms underlying fetal established that the key regulatory proteins that mediate cell cycle ISCs conversion, we have here identified a restricted genetic block include p53, p21 and p16, among others (reviewed in signature that is present in a subset of tumors that are mostly ref. 39). Notably, cancer cells carrying dysfunctional p53 show the included in the CMS4 but clearly different from that of partial conversion to fetal phenotype but fail to acquire a func- adult ISCs. tional PQL phenotype, as they lose TIC associated with massive From a clinical perspective, uncovering genetic signatures that accumulation of DNA damage, when treated at doses that reduce are predictive of recurrence in a group of patients with uncertain cell growth ~20–30 %. Although it is still possible that p53 projections (stages II and III) will represent a powerful tool for mutated cells can acquire a PQL phenotype at specific CT doses, diagnosis refinement. In this direction, we are currently setting up we found that p53 deletion by itself led to the accumulation of the protocols for early detection of PQL/feISC cells in stages II–III DNA damage and imposed defective clonogenic activity after tumors at diagnosis. As mentioned, anticipating the presence of exposure to comparable levels of 5-FU + Iri. this adverse phenotype in tumors would allow exposing selected Our findings, which are in agreement with the recent groups of patients to alternative therapeutic procedures that could demonstration that tumor cells that persist after prolonged CT be refined with the discovery of the mechanisms imposing fetal acquires an embryonic-like and quiescent state that facilitates SC conversion in cancer. Moreover, our data alert about potential therapeutic resistance14 will also help to clarify the functional negative effects of neoadjuvant treatment and indicate the pos- contribution of YAP1 as a driver of fetal conversion and driver or sibility of using nuclear YAP1 or specific fetal ISC markers to suppressor of metastasis (reviewed in ref. 40). We propose that identify patients that would benefit from future YAP1-based functional p53 through p21 allows cells to recover from DNA therapeutic strategies. damage under specific conditions, whereas p53 mutant cells continue proliferating leading to irreparable damage and apop- Methods totic death. This possibility would reconcile our observations with Study design. The goal of this study was to determine the impact of sublethal CT a previous publication indicating that fetal reprograming of in colorectal cancer cells. A study of sublethal doses of CT was performed in several intestinal cancer cells induced by YAP1 led to tumor and PDOs and human cell lines. We identified genetic signatures induced by CT that were evaluated in public colorectal tumor databases for their prognosis value. metastasis suppression in the Apc−/−; KrasG12D; p53−/− Numbers of experiments, biological replicates, and sample sizes are outlined in the murine cancer model16. Specifically, our results indicate that fetal figure legends. ISC conversion is not induced by p53 transcriptionally, but p53 protects fetal-converted cells from massive DNA damage and loss Cell lines. CRC cell lines HCT116 and Ls174T (KRAS mutated and TP53 WT), of TIC and metastatic potential after sublethal CT. This cellular SW480 (KRAS and TP53 mutated), and HT29 (BRAF and TP53 mutated) were response represents, in fact, a double-edged sword since it can obtained from the American Type Culture Collection (ATCC, USA). All cells were grown in Dulbecco’s modified Eagle’s medium (Invitrogen) plus 10% fetal bovine impose specific outcomes depending on the TP53 status of cancer serum (Biological Industries) and were maintained in a 5% CO2 incubator at 37 °C. cells. In this sense, 5-FU treatment induces cell dormancy and 5-FU + Iri. concentrations that reduced 30% of each cell growth were as follows: epithelial-to mesenchyme transition in lung cancer cells, asso- HCT116, 0.01 µg/mL 5-FU and 0.004 µg/mL Iri.; Ls174T, 0.025 µg/mL 5-FU and ciated with p53 accumulation41. Further experiments genetically 0.01 µg/mL Iri.; SW480, 0.28 µg/mL 5-FU and 0.11 µg/mL Iri.; HT29, 0.33 µg/mL 5-FU and 0.13 µg/mL Iri. deleting YAP1 in TP53 WT cells and preclinical assays using YAP1 inhibitors are required to demonstrate the possibility of designing specific protocols to treat fetal-converted tumors. Human colorectal tumors. Formalin-fixed, paraffin-embedded tissue blocks of gastrointestinal tumor samples, from patients at diagnosis and after neoadjuvant Our findings linking fetal ISC conversion with poor CRC therapy at the time of surgery, were obtained from Parc de Salut Mar Biobank prognosis are partially opposite to the idea that adult ISC sig- (MARBiobank, Barcelona). Samples were retrieved under informed consent and natures are indicative of tumor malignancy3. Nevertheless, it has approval of the Clinical Research Ethics Committee-Parc de Salut Mar (CEIC- been recently demonstrated that Lgr5 and other adult ISC mar- PSMAR) according to Spanish ethical regulations and the guidelines of the Declaration of Helsinki. Patient identity for pathological specimens remained kers are temporarily lost from cells seeding metastases, and anonymous in the context of this study. Patient data was collected and treatment subsequently recovered (due to cellular plasticity) to allow regimens were standard and adjusted to patient and tumor characteristics (see metastasis establishment42. Moreover, most data indicating the Supplementary Tables S2 and S6). IHC analyses were performed as requirement for adult ISCs in metastasis seeding were obtained described below. on TP53 -deficient tumor cells16,43. Recently, it was found that Lgr5+ cells can differentially contribute to tumor progression and Animal studies. For tumor-initiating/metastasis assays we followed two approa- ches: 1) Intracardiac injection of 40,000 control, IC20 or IC30-treated PDO5 cells authors showed higher Lgr5+ frequency in mutant TP5344, thus carrying a luciferase reporter to NSG mice in two independent experiments (strain: opening the possibility that Lgr5+ ISCs dependence in cancer is ANB//NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ; ten-week-old males). To determine 12 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE tumor load, animals were anesthetized and injected with 100 µl of substrate Immunofluorescence analysis. For tissues and PDOs, the same protocol as IHC D-luciferin at 15 mg/ml intraorbital. Bioluminescent images were taken at day 0 was followed. However, the samples were developed with Tyramide Signal and every week in the IVIS Lumina III In Vivo Imaging System (PerkinElmer) with Amplification System (TSA) and mounted in DAPI Fluoromount-G. Images were 2 min exposure. Metastasis initiation differences were determined by using the two- taken in an SP5 upright confocal microscope (Leica). way ANOVA test. Quantification was done using Living Image® software (Perki- nElmer). 2) Equivalent amounts of disaggregated PDO cells, previously treated as Hematoxylin and eosin staining. Previously de-paraffinized sections were incu- indicated, were implanted as orthoxenografts athymic nude mice (strain: Hsd:A- bated with hematoxylin 30 s, tap water 5 min, 80% ethanol 0.15% HCl 30 s, water thymic Nude-Foxn1nu; 5–7-week-old males). Tumor growth was determined by 30 s, 30% ammonia water (NH3(aq)) 30 s, water 30 s, 96% ethanol 5 min, eosin 3 s, palpation, and animals were sacrificed when controls developed tumors of around and absolute ethanol 1 min. Samples were dehydrated, and mounted in DPX, and 2 cm in diameter. Procedures involving living animals were conducted under images were obtained with an Olympus BX61 microscope. pathogen-free conditions and according to guidelines from the Animal Care Committee at the Generalitat de Catalunya. The Committee for Animal Experi- mentation at the Institute of Biomedical Research of Bellvitge (Barcelona) approved FISH. Fluorescent in situ hybridization (FISH) analyses from untreated and IC30- these studies. treated PDOs were performed using commercial probes (Abbott Molecular Inc, Des Plaines, IL, USA), one including the centromeric alfa-satellite region specific for chromosome 8, and a second one containing locus-specific probes from the PDO and culture conditions. Human colorectal tumors were obtained from Parc long arm of chromosome 13 and 21. In brief, we performed a cytospin to con- de Salut MAR Biobank (MARbiobank) and IdiPAZ Biobank, integrated into the centrate nuclei in the FISH slide. Slides were pre-treated with pepsin for 5 min at Spanish Hospital Biobanks Network (RetBioH; www.redbiobancos.es). Written 37 °C. Samples and probes were co-denaturated at 80 °C for five minutes and informed consent was obtained from all participants and protocols were approved hybridized overnight at 37 °C in a hot plate (Hybrite chamber, Abbot Molecular by Hospital del Mar’ Ethics Committee (approval number 2019/8595/I), the Inc.). Post-hybridization washes were performed at 73 °C in 2× sodium salt citrate Spanish regulations, and the Helsinki declaration’s Guide. For PDOs generation, buffer (SSC) and at room temperature in 2× SSC, 0.1% NP-40 solution. Samples primary or xenografted human colorectal tumors were disaggregated in 1.5 mg/mL were counterstained with 4,6-diamino-2-phenilindole (DAPI)(Abbott Molecular collagenase II and 20 μg/mL hyaluronidase after 40 min of incubation at 37 °C, Inc, Des Plaines, IL, USA). Results were analyzed in a fluorescence microscope filtered in 100 μm cell strainer, and seeded in 50 μL Matrigel in 24-well plates. After (Olympus, BX51) using the Cytovision software (Applied Imaging, Santa Clara, polymerization, 450 μL of complete medium was added (DMEM/F12 plus peni- CA). A minimum of 50 nuclei per case was analyzed. cillin (100 U/mL) and streptomycin (100 μg/mL), 100 μg/mL Primocin, 1× N2 and B27, 10 mM Nicotinamide; 1.25 mM N-Acetyl-L-cysteine, 100 ng/mL Noggin and 100 ng/mL R-spondin-1, 10 μM Y-27632, 10 nM PGE2, 3 μM SB202190, 0.5 μM A- Comet assay. Comet assays were performed using Comet Assay Kit following 8301, 50 ng/mL EGF and 10 nM Gastrin I). Tumor spheres were collected and manufacturer’s instructions. Pictures were taken using a Nikon Eclipse Ni-E epi- digested with an adequate amount of trypsin to single cells and re-plated in culture. fluorescence microscope and tail moment was calculated using the OPENCOMET Cultures were maintained at 37 °C, 5% CO2 and medium changed every week. plugin for Fiji. PDOs were expanded by serial passaging and kept frozen in liquid Nitrogen for being used in subsequent experiments. Mutations identified in the PDOs are listed Annexin V binding assay. Annexin V binding was determined by flow cytometry in Supplementary Table S1. using the standard Annexin V Apoptosis Detection Kit APC. Single cells of treated PDOs with indicated combinations of 5-FU + Iri. were obtained and stained PDO viability assays. 600 single PDO cells were plated in 96-well plates in 10 μL according to the manufacturer’s instructions, with Propidium Iodide staining for Matrigel with 100 μL of complete medium. After 6 days in culture, growing PDOs the DNA content. The cells were analyzed in the Fortessa analyzer. were treated with combinations of 5-FU + Iri. for 72 h at the concentrations that reduce 20 and 30% of the cell growth (IC20 and IC30, respectively), which are Cell senescence assays. Cell senescence was identified by the presence of SA-β- specific for each PDO as described in Supplementary Table S1. After 72 h of galactosidase activity using two different approaches. On one hand, staining for treatment, we changed to a fresh medium and measured the cell viability after SA-β-galactosidase activity in cultured cells was carried out using the Senescence β- 3 days, 1 week, and 2 weeks using the CellTiter-Glo 3D Cell Viability Assay Galactosidase Staining Kit. Briefly, PDOs were seeded in 24-well plates (3000 cells following manufacturer’s instructions in an Orion II multiplate luminometer. per well). After 6 days, PDOs were treated with combinations of 5-FU + Iri. for Images were obtained with an Olympus BX61 microscope at the indicated time 72 h and were subsequently stained with the β-Galactosidase Staining Solution for points and the diameter of at least 70 tumoroids per condition was determined 2 h, according to the manufacturer’s instructions. Sections embedded in paraffin using Adobe Photoshop. For dose–response curves, PDOs were plated in 96-well were counterstained with Fast Red for nuclei visualization. Images were obtained plates in Matrigel and after 6 days in culture were treated with combinations of with an Olympus BX61 microscope. On the other hand, SA-β-galactosidase activity 5-FU and Irinotecan. Following 72 h of treatment, we changed to fresh medium was addressed by flow cytometry using the Cell Event Senescence Green Flow and treated with increasing concentrations of either 5-FU + Iri., dasatinib, verte- Cytometry Assay Kit following the manufacturer’s instructions, and analyzed in the porfin or combinations for 72 h at the indicated concentrations. Cell viability was LSR II analyzer. determined as described above. Cell cycle analysis. Cell cycle was determined by flow cytometry using the stan- TIC assay. For TIC assay in vitro, 300 or 600 single PDO cells were plated in 96- dard APC BrdU Flow Kit. Briefly, treated PDOs with combinations of 5-FU + Iri., well plates in 10 μL Matrigel. After 11 days in culture, the number of PDOs in each as indicated, were stained with bromodeoxyuridine (BrdU) for 24 h. Single cells well was counted, photographs were taken for PDO diameter determination and were obtained and processed according to the manufacturer’s instructions, with cell viability was measured. DAPI staining for the DNA content. The cells were analyzed in the LSR II analyzer. PDOs infection. hFLiG plasmid was used for in vivo detection of metastasis, Cell lysis and Western Blot. Treated PDOs were lysed for 20 min on ice in 300 μL H2BeGFP plasmid was used for flow cytometry experiments and lentiCRISPR v2 of PBS plus 0.5% Triton X-100, 1 mM EDTA, 100 mM NA-orthovanadate, 0.2 mM was used for knock-out experiments. Three sgRNA against TP53 and YAP1 gene phenyl-methylsulfonyl fluoride, and complete protease and phosphatase inhibitor were designed using Benchling (Supplementary Table S5). Lentiviral production cocktails. Lysates were analyzed by western blotting using standard was performed by transfecting HEK293T cells the lentiviral vectors and the plas- SDS–polyacrylamide gel electrophoresis (SDS-PAGE) techniques. In brief, protein mid of interest. One day after transfection, the medium was changed, and viral samples were boiled in Laemmli buffer, run in polyacrylamide gels, and transferred particles were collected 24 h later and then concentrated using Lenti-X Con- onto polyvinylidene difluoride (PVDF) membranes. The membranes were incu- centrator. PDOs were infected by resuspending single cells in concentrated virus bated with the appropriate primary antibodies overnight at 4 °C, washed, and diluted in a complete medium, centrifuged for 1 h at 650 rcf, and incubated for 5 h incubated with specific secondary horseradish peroxidase–linked antibodies. Per- at 37 °C. Cells were then washed in a complete culture medium and seeded as oxidase activity was visualized using the enhanced chemiluminescence reagent and described above. autoradiography films. Immunohistochemical staining. Paraffin blocks were obtained from tissues and RT-qPCR analysis. Total RNA from treated PDOs was extracted with the RNeasy PDOs, the previous fixation in 4% formaldehyde overnight at room temperature. Micro Kit, and cDNA was produced with the RT-First Strand cDNA Synthesis Kit. Paraffin-embedded sections of 4 μm, for tissues, and 2.5 μm, for PDOs, were de- RT-qPCR was performed in LightCycler 480 system using SYBR Green I Master paraffinized, rehydrated and endogenous peroxidase activity was quenched Kit. Samples were normalized relative to the housekeeping genes TBP and/or (20 min, 1.5% H2O2). EDTA or citrate‐based antigen retrieval was used depending HPRT1. Primers used for RT-qPCR are listed in Supplementary Table S5. on the primary antibody used. All primary antibodies were diluted in PBS con- taining 0.05% BSA, incubated overnight at 4 °C, and developed with the Envision+ System HRP Labeled Polymer anti-Rabbit or anti-Mouse and 3,3′-diaminobenzi- RNA-sequencing and analysis. Total RNA from untreated and treated PDOs was dine (DAB). Samples were mounted in DPX and images were obtained with an extracted using RNeasy Micro Kit. The RNA concentration and integrity were Olympus BX61 microscope. determined using Agilent Bioanalyzer [Agilent Technologies]. Libraries were NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 13 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 prepared at the Genomics Unit of PRBB (Barcelona, Spain) using standard pro- Each experiment shown in the manuscript has been repeated at least twice. tocols, and cDNA was sequenced using Illumina HiSeq platform, obtaining ~45–64 Combinations of 5-FU + Iri. treatment has been checked for an appropriate IC20 million 50-bp paired‐end reads per sample. Adapter sequences were trimmed with and IC30 effect in every experiment, by cell viability assay. Bioinformatic analyses Trim Galore. Sequences were filtered by quality (Q > 30) and length (>20 bp). were performed as indicated above. Filtered reads were mapped against the latest release of the human reference genome (hg38) using default parameters of TopHat (v.2.1.1)1 and expressed Reporting summary. Further information on research design is available in the Nature transcripts were then assembled. High-quality alignments were fed to HTSeq Research Reporting Summary linked to this article. (v.0.9.1)2 to estimate the normalized counts of each expressed gene. Differentially expressed genes between conditions (considering IC20-, IC30- treated PDO5, and IC30-treated PDO66 as treated conditions and untreated samples from PD05 and PDO66 as control conditions) from RNA-seq data were Data availability explored using DESeq2 R package (v.1.30.1)45 and the removeBatchEffect from the RNA sequencing and ChIP-sequencing data have been deposited in NCBI’s Gene limma package (v.3.46.0) was used to correct technical batch effects. Adjusted P Expression and are accessible through GEO Series accession no. GSE155354 and no. values for multiple comparisons were calculated applying the Benjamini-Hochberg GSE164161 respectively. Public data sets used in this study are also accessible through correction (False Discovery Rate) (see Supplementary Data 1). Plots were done in GEO Series accession no. GSE39582 for Marisa data set and GSE14333 for Jorissen data R. Expression heatmaps were generated using the heatmaply and pheatmap set; and for TCGA data set through the TCGA portal [http://www.tcgaportal.org/]. In packages in R46. GSEA was performed with described gene sets using gene set this study, data from these data sets have been obtained using CANCERTOOL platform permutations (n = 1000) for the assessment of significance and signal-to-noise [http://genomics.cicbiogune.es/CANCERTOOL/]. Data from the experiments are metric for ranking genes. RNA-sequencing data for PDO5 and PDO66 are deposited at the GEO database with accession number GSE155354. provided in the Supplementary Information. Source data are provided with this paper. ChIP-sequencing and analysis. Formaldehyde cross-linked cell extracts of IC20- treated PDO5 were sonicated, and chromatin fractions were incubated for 16 h Code availability with anti-p53 [abcam ab 1101] antibody in RIPA buffer and then precipitated with No new codes were used in this study. protein A/G-sepharose [GE Healthcare, Refs. 17-0618-01 and 17-0780-01]. Crosslinkage was reversed, and 6–10 ng of precipitated chromatin was directly sequenced in the genomics facility of Parc de Recerca Biomèdica de Barcelona Received: 30 March 2021; Accepted: 29 April 2022; (PRBB) using Illumina® HiSeq platform. Raw single-end 50-bp sequences were filtered by quality (Q > 30) and length (length > 20 bp) with Trim Galore4. Filtered sequences were aligned against the reference genome (hg38) with Bowtie25. MACS2 software6 was run first for each replicate using unique alignments (q value < 0.1). Peak annotation was performed with ChIPseeker package7 and peak visualization was done with Integrative Genomics Viewer (IGV). ChIP-sequencing References data are deposited at the GEO database with accession number GSE164161. 1. Brenner, H., Kloor, M. & Pox, C. P. Colorectal cancer. Lancet https://doi.org/ 10.1016/S0140-6736(13)61649-9 (2014). ChIP-qPCR. Untreated and IC20-treated PDOs were subjected to ChIP following 2. Batlle, E. & Clevers, H. Cancer stem cells revisited. Nat. Med. https://doi.org/ standard procedures. Briefly, PDO cells were extracted with formaldehyde cross- 10.1038/nm.4409 (2017). linked for 10 min at room temperature and lysed for 20 min on ice with 500 μL of 3. Merlos-Suárez, A. et al. The intestinal stemcell signature identifies colorectal H2O plus 10 mM Tris-HCl pH 8.0, 0.25% Triton X-100, 10 mM EDTA, 0.5 mM cancer stem cells and predicts disease relapse. Cell Stem Cell 8, 511–524 (2011). EGTA, 20 mM β-glycerol-phosphate, 100 mM NA-orthovanadate, 10 mM NaBu- 4. Galmarini, F. C., Galmarini, C. M., Sarchi, M. I., Abulafia, J. & Galmarini, D. tyrate and complete protease inhibitor cocktail. The supernatants were sonicated, Heterogeneous distribution of tumor blood supply affects the response to centrifuged at 13,000 rpm for 15 min, and supernatants were incubated overnight chemotherapy in patients with head and neck cancer. Microcirculation 7, with anti-p53 antibody in RIPA buffer. Precipitates were captured with 35 mL of 405–410 (2000). protein A-Sepharose, extensively washed, and analyzed by ChIP-qPCR. The pri- 5. Azzi, S., Hebda, J. K. & Gavard, J. Vascular permeability and drug delivery in mers used are listed in Supplementary Table S5. Inputs were used to normalize the cancers. Front. Oncol. 3, 211 (2013). ChIP-qPCR and samples were compared to control IgGs. 6. Colomer, C. et al. IKKα kinase regulates the DNA damage response and drives chemo-resistance in cancer. Mol. Cell 75, 669–682.e5 (2019). 7. Cánovas, B. et al. Targeting p38α increases DNA damage, chromosome Description of the patient gene expression data sets. Transcriptomic and instability, and the anti-tumoral response to taxanes in breast cancer cells. available clinical data sets from colorectal cancer were downloaded from the open- Cancer Cell https://doi.org/10.1016/j.ccell.2018.04.010 (2018). access resource CANCERTOOL. For CRC we used the Marisa (GSE39582) data set, 8. He, S. & Sharpless, N. E. Senescence in health and disease. Cell https://doi.org/ which includes expression and clinical data for 566 patients with CRC and 19 non- tumoral colorectal mucosa, the Jorissen (GSE14333) data set and the TCGA data 10.1016/j.cell.2017.05.015 (2017). set with expression and clinical data of 226 and 329 CRC patients, respectively. 9. Faget, D. V., Ren, Q. & Stewart, S. A. Unmasking senescence: context- dependent effects of SASP in cancer. Nat. Rev. Cancer https://doi.org/10.1038/ s41568-019-0156-2 (2019). Association of the signatures with clinical outcome. Association of the 10. Milanovic, M. et al. Senescence-associated reprogramming promotes cancer expression of the signature with relapse was assessed in the cancer transcriptomic stemness. Nature https://doi.org/10.1038/nature25167 (2018). data sets using Kaplan–Meier estimates and Cox proportional hazard models. A 11. Schosserer, M., Grillari, J. & Breitenbach, M. The dual role of cellular standard log-rank test was applied to assess significance between groups. This test senescence in developing tumors and their response to cancer therapy. Front. was selected because it assumes the randomness of the possible censorship. All the Oncol. https://doi.org/10.3389/fonc.2017.00278 (2017). survival analyses and graphs were performed with R using the survival (v.3.2-3) 12. Brown, J. A. et al. TGF-β-induced quiescence mediates chemoresistance of and survimer (v.0.4.8) packages and a p value<0.05 was considered statistically tumor-propagating cells in squamous cell carcinoma. Cell Stem Cell https:// significant (see Supplementary Table S3). doi.org/10.1016/j.stem.2017.10.001 (2017). 13. Liau, B. B. et al. Adaptive chromatin remodeling drives glioblastoma stem cell Signature definition. To generate the feISC signatures, we selected genes with log2 plasticity and drug tolerance. Cell Stem Cell 20, 233–246.e7 (2017). Fold Change (FC) TreatedvsControl > 0 and FetalvsAdult30 >0 in the case of the 14. Rehman, S. K. et al. Colorectal cancer cells enter a diapause-like DTP state to 28up-feISC and log2FC TreatedvsControl < 0 and FetalvsAdult < 0 in the case of survive chemotherapy. Cell https://doi.org/10.1016/j.cell.2020.11.018 (2021). the 8down-feISC. Next, we used the Marisa data set to perform expression cor- 15. Touil, Y. et al. Colon cancer cells escape 5FU chemotherapy-induced cell relation matrices for the selected expression gene pairs using the corrplot package death by entering stemness and quiescence associated with the c-Yes/YAP (v.0.84). Correlations were considered statistically significant when the Pearson axis. Clin. Cancer Res. 20, 837–846 (2014). correlation coefficient corresponded to a p value below 0.05. Clusters of genes were 16. Cheung, P. et al. Regenerative reprogramming of the intestinal stem cell state selected when the absolute value for the Pearson correlation coefficient was via hippo signaling suppresses metastatic colorectal cancer. Cell Stem Cell 27, above 0.1. 590–604.e9 (2020). 17. Lukin, D. J., Carvajal, L. A., Liu, W. J., Resnick-Silverman, L. & Manfredi, J. J. Quantification and Statistical analysis. Statistical parameters, including the p53 promotes cell survival due to the reversibility of its cell-cycle checkpoints. number of events quantified, standard deviation, and statistical significance, are Mol. Cancer Res. https://doi.org/10.1158/1541-7786.MCR-14-0177 (2015). reported in the figures and in the figure legends. Statistical analysis has been 18. Saleh, T., Tyutyunyk-Massey, L. & Gewirtz, D. A. Tumor cell escape from performed using GraphPad Prism 6 software, and P < 0.05 is considered significant. therapy-induced senescence as a model of disease recurrence after dormancy. Two-sided Student’s t test was used to compare differences between two groups. Cancer Res. https://doi.org/10.1158/0008-5472.CAN-18-3437 (2019). 14 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-022-30382-9 ARTICLE 19. Xu, M. et al. Senolytics improve physical function and increase lifespan in old 46. Galili, T., O’Callaghan, A., Sidi, J. & Sievert, C. Heatmaply: an R package for age. Nat. Med. https://doi.org/10.1038/s41591-018-0092-9 (2018). creating interactive cluster heatmaps for online publishing. Bioinformatics 20. Puig, I. et al. TET2 controls chemoresistant slow-cycling cancer cell survival https://doi.org/10.1093/bioinformatics/btx657 (2018). and tumor recurrence. J. Clin. Invest. https://doi.org/10.1172/JCI96393 (2018). 21. Cho, Y. H. et al. 5-FU promotes stemness of colorectal cancer via p53- mediated WNT/β-catenin pathway activation. Nat. Commun. https://doi.org/ Acknowledgements 10.1038/s41467-020-19173-2 (2020). We want to thank the Bigas’ and Espinosa’s lab members for constructive discussions 22. Mani, S. A. et al. The epithelial-mesenchymal transition generates cells with and suggestions and technical support. We thank our patients for their generosity and to properties of stem cells. Cell https://doi.org/10.1016/j.cell.2008.03.027 (2008). MARbiobank and the IdiPAZ Biobank integrated with the Spanish Hospital Biobanks 23. Liu, M. et al. The canonical NF-κB pathway governs mammary tumorigenesis Network (RetBioH; www.redbiobancos.es). This work was funded by Instituto de Salud in transgenic mice and tumor stem cell expansion. Cancer Res. https://doi.org/ Carlos III (ISCIII) and co-funded by the European Union (PI16/00437 and PI19/00013), 10.1158/0008-5472.CAN-10-0732 (2010). Generalitat de Catalunya 2017SGR135, PID2019-104867RB-I00/AEI/10.13039/ 24. Rajasekhar, V. K., Studer, L., Gerald, W., Socci, N. D. & Scher, H. I. Tumour- 501100011033 funded by the Agencia Estatal de Investigación and the “Xarxa de Bancs initiating stem-like cells in human prostate cancer exhibit increased NF-κB de tumors sponsored by Pla Director d’Oncologia de Catalunya (XBTC), Fundación signalling. Nat. Commun. https://doi.org/10.1038/ncomms1159 (2011). HNA to A.V. research team in the development of orthoxenografts/PDOX., Instituto de 25. Kagoya, Y. et al. Positive feedback between NF-κB and TNF-α promotes Salud Carlos III-Fondo Europeo de Desarrollo Regional (CIBERONC; CB16/12/00244, leukemia-initiating cell capacity. J. Clin. Invest. https://doi.org/10.1172/ CB16/12/00241 and CB16/12/00273). IdiPAZ Biobank is supported by Instituto de Salud JCI68101 (2014). Carlos III, Spanish Health Ministry (Grant RD09/0076/00073), and Farmaindustria 26. Nusse, Y. M. et al. Parasitic helminths induce fetal-like reversion in the through the Cooperation Program in Clinical and Translational Research of the Com- intestinal stem cell niche. Nature https://doi.org/10.1038/s41586-018-0257-1 munity of Madrid. LS is supported by AGAUR (2018 FI_B 00088/2020 FI_B2 00150). (2018). D.A-V was funded by FI20/00130 and Y.G. and T.L-J by a contract from CIBERONC 27. Rubin, S. M., Sage, J. & Skotheim, J. M. Integrating old and new paradigms of (ISCIII-Feder) and Asociación Española Contra el Cancer (AECC). T.C-T was funded by G1/S control. Mol. Cell 80, 183–192 (2020). the Instituto de Salud Carlos III-FSE (MS17/ 00037; PI18/00014). A.V was supported by 28. Muñoz, J. et al. The Lgr5 intestinal stem cell signature: robust expression of grant PI19/01320 funded by Instituto de Salud Carlos III (ISCIII) and co-funded by the proposed quiescent ′+4′ cell markers. EMBO J. https://doi.org/10.1038/emboj. European Union. 2012.166 (2012). 29. Yui, S. et al. YAP/TAZ-dependent reprogramming of colonic epithelium links Author contributions ECM remodeling to tissue regeneration. Cell Stem Cell https://doi.org/10.1016/ Conceptualization: A.B., L.E., biochemical assays, in vitro and in vivo experiments: A.Vi., j.stem.2017.11.001 (2018). M.G., M.S., R.G.-R., M.M.-I., I.S., D.A.-V., J.A.-M., J.G., C.R. Experiments, investigation 30. Mustata, R. C. et al. Identification of Lgr5-independent spheroid-generating and evaluation of results: A.Ba., L.S., T.L.-J., A.Ve., T.C.-T., A.M., L.E. Big data analysis progenitors of the mouse fetal intestinal epithelium. Cell Rep. https://doi.org/ and statistical analysis: T.L.-J., Y.G., E.L.-A., F.T. Clinicopathological characterization of 10.1016/j.celrep.2013.09.005 (2013). human tumors: M.G., R.S., J.L., M.I., C.M. Clinical advice: M.G., B.B., R.S., C.M. 31. Gregorieff, A., Liu, Y., Inanlou, M. R., Khomchuk, Y. & Wrana, J. L. Yap- Writing–original drat: L.E., A.Bi. Writing–review & editing: all authors dependent reprogramming of Lgr5+ stem cells drives intestinal regeneration and cancer. Nature 526, 715–718 (2015). 32. Marisa, L. et al. Gene expression classification of colon cancer into molecular Competing interests subtypes: characterization, validation, and prognostic value. PLoS Med. Laura Solé, Teresa Lobo-Jarne, Marta Guix, Beatriz Bellosillo, Mar Iglesias, Anna Bigas, https://doi.org/10.1371/journal.pmed.1001453 (2013). and Lluís Espinosa have a pending patent application entitled Genomic predictor of 33. Jorissen, R. N. et al. Metastasis-associated gene expression changes predict outcome in cancer with number PCT/EP2022/058503, related to the fetal signature as a poor outcomes in patients with Dukes stage B and C colorectal cancer. Clin. prognostic biomarker in colorectal cancer. The authors have no additional financial Cancer Res. https://doi.org/10.1158/1078-0432.CCR-09-1431 (2009). interests and the remaining authors declare no competing interests. 34. Cortazar, A. R. et al. Cancertool: a visualization and representation interface to exploit cancer datasets. Cancer Res. https://doi.org/10.1158/0008-5472.CAN- 18-1669 (2018). Additional information Supplementary information The online version contains supplementary material 35. Guinney, J. et al. The consensus molecular subtypes of colorectal cancer. Nat. available at https://doi.org/10.1038/s41467-022-30382-9. Med. https://doi.org/10.1038/nm.3967 (2015). 36. Lee, H. O. et al. Lineage-dependent gene expression programs influence the Correspondence and requests for materials should be addressed to Anna Bigas or Lluís immune landscape of colorectal cancer. Nat. Genet. https://doi.org/10.1038/ Espinosa. s41588-020-0636-z (2020). 37. Pefani, D. E. et al. TGF-β Targets the Hippo Pathway Scaffold RASSF1A to Peer review information Nature Communications thanks Kevin Myant, Antonio Biondi, Facilitate YAP/SMAD2 Nuclear Translocation. Mol. Cell https://doi.org/10. and the other anonymous reviewer(s) for their contribution to the peer review of this 1016/j.molcel.2016.05.012 (2016). work. Peer review reports are available. 38. Kreso, A. et al. Variable clonal repopulation dynamics influence chemotherapy response in colorectal cancer. Science. https://doi.org/10.1126/ Reprints and permission information is available at http://www.nature.com/reprints science.1227670 (2013). 39. Weinberg, R. A. The retinoblastoma protein and cell cycle control. Cell https:// Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in doi.org/10.1016/0092-8674(95)90385-2 (1995). published maps and institutional affiliations. 40. Warren, J. S. A., Xiao, Y. & Lamar, J. M. YAP/TAZ activation as a target for treating metastatic cancer. Cancers https://doi.org/10.3390/cancers10040115 (2018). 41. Dai, Y. et al. Activation of anaphase-promoting complex by p53 induces a Open Access This article is licensed under a Creative Commons state of dormancy in cancer cells against chemotherapeutic stress. Oncotarget Attribution 4.0 International License, which permits use, sharing, https://doi.org/10.18632/oncotarget.8172 (2016). adaptation, distribution and reproduction in any medium or format, as long as you give 42. Fumagalli, A. et al. Plasticity of Lgr5-negative cancer cells drives metastasis in appropriate credit to the original author(s) and the source, provide a link to the Creative colorectal cancer. Cell Stem Cell https://doi.org/10.1016/j.stem.2020.02.008 Commons license, and indicate if changes were made. The images or other third party (2020). material in this article are included in the article’s Creative Commons license, unless 43. De Sousa E Melo, F. et al. A distinct role for Lgr5 + stem cells in primary and indicated otherwise in a credit line to the material. If material is not included in the metastatic colon cancer. Nature https://doi.org/10.1038/nature21713 (2017). article’s Creative Commons license and your intended use is not permitted by statutory 44. Morral, C. et al. Zonation of ribosomal DNA transcription defines a stem cell regulation or exceeds the permitted use, you will need to obtain permission directly from hierarchy in colorectal cancer. Cell Stem Cell https://doi.org/10.1016/j.stem. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ 2020.04.012 (2020). licenses/by/4.0/. 45. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. https://doi.org/10. 1186/s13059-014-0550-8 (2014). © The Author(s) 2022 NATURE COMMUNICATIONS | (2022)13:2866 | https://doi.org/10.1038/s41467-022-30382-9 | www.nature.com/naturecommunications 15