40% of colorectal cancer (CRC) patients undergoing curative resection of the primary tumor will develop metastases in the following years 1 . Therapies to prevent disease relapse remain an unmet medical need. Here we uncover the identity and features of the residual tumor cells responsible for CRC relapse. Analysis of single-cell transcriptomes of CRC patient samples revealed that the majority of poor prognosis genes are expressed by a unique tumor cell population that we named High Relapse Cells (HRCs). We established a human-like mouse model of microsatellite stable CRC that undergoes metastatic relapse following surgical resection of the primary tumor. Residual HRCs occult in mouse livers after primary CRC surgery gave rise to multiple cell types over time, including Lgr5+ stemlike tumor cells 2-4 , and caused overt metastatic disease. Using Emp1 (epithelial membrane Competitiveness (MINECO). HH is a Miguel Servet (CP14/00229) researcher funded by the

The standard of care for advanced colorectal cancer (CRC) includes treatment with chemotherapeutic drugs that target the cell proliferation machinery 1 . In CRC patients with overt metastases, chemotherapy initially halts tumor growth but, almost inevitably, disease progresses after some cycles of treatment. Adjuvant chemotherapy is also administered to eliminate minimal residual disease, yet it only diminishes the risk of relapse by 10-25% 2 . Previous studies have shown that patient-derived organoids predict responses to chemotherapy 3-6 . Therefore, we used them as models to investigate the mechanisms behind the limited benefit of these treatments. Whereas CRC organoids expand from highly proliferative Lgr5+ tumor cells, we discovered that lack of optimal stem cell growth conditions specifies a latent Lgr5+ cell population. These cells expressed the gene Mex3a, were largely insensitive to chemotherapy and regenerated the organoid culture after treatment. In mouse models of metastatic latency, Mex3a+ cells contributed marginally to metastatic outgrowth. However, after chemotherapy treatment, Mex3a+ cells produced large cell clones that regenerated metastatic disease. Using lineage-tracing analysis combined with single cell profiling, we showed that drug-tolerant persister Mex3a+ cells downregulate the WNT/Lgr5+ stem cell program immediately after chemotherapy and adopt a transient regenerative state

Patient-derived organoids (PDOs) have demonstrated predictive value in prospective clinical trials supporting selection of personalized treatments. Because PDOs retain the organization and physiological functions of their source tissue, PDO biobanks could also be an ideal substrate to screen for novel therapeutic interventions. Here we describe a large-scale functional screen of dual targeting bispecific antibodies (bAbs) on a colorectal cancer (CRC) PDO biobank to target their dependency on cancer stem cells. A novel drug discovery pipeline was assembled where therapeutic bAb panels generated against WNT and receptor tyrosine kinases (RTK) targets were functionally evaluated by high content imaging to capture the complexity of PDO responses across a wide range of different CRCs and paired normal colonic mucosa samples. Our strategy resulted in the generation of MCLA-158, a bAb that specifically triggers EGFR degradation in LGR5+ cancer stem cells but shows minimal toxicity towards normal LGR5+ colon stem cells. MCLA-158 exhibits unique therapeutic properties such as potent growth inhibition of KRAS mutant CRCs, blockade of metastasis initiation and suppression of tumor outgrowth in preclinical models of different cancer types.