Apc-mutant cells act as supercompetitors in intestinal tumour initiation

van Neerven, S.M., de Groot, N.E., Nijman, L.E. et al. Apc-mutant cells act as supercompetitors in intestinal tumour initiation. Nature 594, 436–441 (2021). https://doi.org/10.1038/s41586-021-03558-4

Sanne M van Neerven^1,2, Nina E de Groot^1,2, Lisanne E Nijman^1,2, Brendon P Scicluna^3,4, Milou S van Driel^1,2, Maria C Lecca^1,2, Daniël O Warmerdam^1,2,5, Vaishali Kakkar^1,2, Leandro F Moreno^1,2, Felipe A Vieira Braga^1,2, Delano R Sanches^1,2, Prashanthi Ramesh^1,2, Sanne ten Hoorn^1,2, Arthur S Aelvoet⁶, Marouska F van Boxel^1,2, Lianne Koens⁷, Przemek M Krawczyk⁸, Jan Koster⁹, Evelien Dekker⁶, Jan Paul Medema^1,2, Douglas J Winton¹⁰, Maarten F Bijlsma^1,2, Edward Morrissey¹¹, Nicolas Léveillé^1,2 & Louis Vermeulen^1,2

Abstract

A delicate equilibrium of Wnt agonists and antagonists in the niche is critical to maintain the intestinal stem cell (ISC) compartment, as it accommodates the rapid renewal of the gut lining. Disruption of this balance by mutations in tumour suppressor gene APC leads to unrestrained Wnt pathway activation and are found in ~80% of all human colon cancers^1,2. Previously, we established that Apc-mutant cells have a competitive advantage over wild type (WT) ISCs3. Consequently, APC-mutant ISCs frequently outcompete all WT stem cells within a crypt, thereby reaching clonal fixation in the tissue and initiating cancer formation. However, it remained unresolved if the increased relative fitness of APC-mutant ISCs involves only cell intrinsic features, or whether APC-mutants are actively involved in the elimination of their WT neighbours. Here we show that APC-mutant ISCs function as bona fide supercompetitors by secreting Wnt antagonists thereby inducing differentiation of neighbouring WT ISCs. Lithium chloride prevented expansion of APC-mutant clones and adenoma formation by rendering WT ISCs insensitive to Wnt antagonists through downstream Wnt activation by GSK3β inhibition. Our work suggests that boosting fitness of healthy cells to limit expansion of pre-malignant clones may be a powerful strategy to limit cancer formation in high-risk individuals.

Main

Colorectal cancer (CRC) formation is a prime example of stepwise cancer development. It is thought that the majority of CRCs are initiated by permanent activation of the Wnt pathway, often through mutations in tumour suppressor gene APC that occur within the stem cell pool⁴. Subsequently, the continuously on-going neutral replacement events between a relatively small number of ISCs residing in the crypt bottom are distorted in the affected crypt, and Apc^-/- ISCs display a positive bias to replace their Apc-proficient neighbours3. As a result, Apc-mutant ISCs and their offspring have an increased probability to fully populate the crypt in which they arise, and initiate tumour formation. Apc mutations induce increased proliferation, prevent cell death, and block differentiation in the intestine^5,6. All these features might contribute to the increased relative fitness of Apc-mutant cells, but given the inherent difficulty of directly targeting the Wnt signalling cascade, to date, these insights have not resulted in more effective therapies or in novel preventive strategies for CRC. Therefore, we set out to study in more detail how Apc-mutant clones exert their competitive advantage over WT ISCs with the aim of identifying signals amenable to pharmacological manipulation. This was achieved by using the combined strengths of in vitro organoid cultures and detailed analyses of in vivo clonal dynamics.

Apc-mutants outcompete wild type cells

We established a co-culture system of WT and Apc^-/- organoids transduced with distinct fluorescent labels (Extended Data Fig. 1a). Whilst the relative surface contribution in WT/WT co-cultures remained constant over time (Figs. 1a, b), Apc^-/- organoids rapidly dominated the co-cultures with WT organoids (Figs. 1c, d), mimicking previous observations in vivo^3,7. This is not simply caused by different proliferation rates of Apc^-/- and WT organoids, instead the WT organoids display reduced expansion rates when co-cultured with Apc^-/- cells, demonstrating that Apc-mutant cells actively suppress growth of WT organoids as determined by surface expansion and cell numbers (Figs. 1e, f). The growth suppressive effect is mediated by secreted factors as conditioned medium (CM) from Apc^-/- organoids, supplemented with fresh growth factors, had a comparable effect (Figs. 1g, h). Typically, Apc^-/- cells arise in a crypt containing Apc^+/- cells following loss of heterozygosity^8,9. While Apc^+/- cells displayed similar expansion rates as Apc^+/+ organoids (Extended Data Figs. 1b-d), we observed Apc^-/- cells also had a growth-reducing effect on Apc^+/- organoids in co-culture (Extended Data Figs. 1e-g). In agreement, Apc^-/- CM also reduced growth of Apc+/- organoids (Extended Data Figs. 1h, i). This indicates that intestinal Apc^-/- cells act as supercompetitors, as they do not passively outcompete their WT and Apc^+/- counterparts but actively subjugate growth of neighbouring cells.

Apc-mutants induce differentiation

To further understand the suppressive influence mediated by Apc-mutant cells on their WT counterpart, we performed transcriptome analysis on WT organoids treated with either WTor Apc^-/- CM (Fig. 2a). WT organoids incubated with Apc^-/- CM displayed features of decreased stemness and increased differentiation as evidenced by reduced expression of Wnt and ISC signatures (Extended Data Figs. 2a-d). These changes mimic the differentiation pattern observed in organoids following R-spondin withdrawal (Figs. 2c, d)¹⁰. The preceding transcriptome-based analyses were confirmed by the fact that WT organoids established from Lgr5-GFP mice that were exposed to Apc^-/- CM, displayed reduced numbers of Lgr5-GFP⁺ cells (Fig. 2e). Furthermore, we confirmed increased MUC2-positive goblet cell numbers in organoids treated with Apc^-/- CM (Fig. 2f). More critically, factors produced by Apc^-/- cells reduced stem cell functionality, as serial passaging of WT organoids exposed to Apc^-/- CM demonstrated markedly reduced clonogenic capacity (Fig. 2g). These findings were corroborated in a human context using organoids established from polypectomies of four individuals with genetically confirmed familial adenomatous polyposis (FAP). CM from APC^-/- organoids from FAP individuals reduced LGR5 expression, increased differentiation markers MUC2 and KRT20, and reduced clonogenicity in WT human organoids (Figs. 2h-j). Together, these data reveal that Apc/APC-mutant murine and human cells secrete factors that actively suppress outgrowth and clonogenicity of WT organoids by promoting differentiation and reducing stem cell numbers.

To determine the nature of the cellular mediators responsible for these observations, we first ruled out the possibility that consumption of metabolites and growth factors by Apc^-/- organoids was involved, as supplementing fresh medium with concentrated CM exerted similar effects (Extended Data Figs. 2e-h). Given the similarity in phenotype to R-spondin withdrawal, and the decrease in Wnt pathway signatures, we speculated that CM from Apc^-/- organoids supressed Wnt activity in WT cells. Indeed Apc^-/- CM significantly reduced recombinant Wnt3a mediated Wnt activation in mouse embryonic fibroblasts (MEFs) that carried a TOP-GFP reporter (Extended Data Figs. 2i-k). Downstream activation of the Wnt pathway by GSK3β inhibitors lithium chloride (LiCl) or CHIR99021 (CHIR) completely abrogated this effect, indicating that pathway inhibition occurs upstream at the ligand/receptor level (Extended Data Figs. 2l, m).

Apc-mutants secrete Wnt antagonists

Transcriptome analysis of murine WT and Apc^-/- organoids revealed that loss of Apc is accompanied by a marked upregulation of several Wnt antagonists, in particular Palmitoleoyl-Protein Carboxylesterase (Notum), Wnt inhibitory factor 1 (Wif1) and Dickkopf-related protein 2 (Dkk2) (Figs. 3a, b). In a time-course experiment we confirmed rapid upregulation of these Wnt antagonists following Apc inactivation in organoid cultures as well as their production in CM (Extended Data Figs. 3a, b). In agreement, we identified upregulation of the same antagonists in murine adenomatous tissue in vivo (Fig. 3c,Extended Data Figs 3c, d). Importantly, a series of partially similar Wnt antagonists were also found upregulated in human APC-deficient organoids (Fig. 3d). In particular NOTUM was also highly upregulated in human FAP derived APC-deficient organoids and adenomas (Figs. 3e, f, Extended Data Figs. 3e-h).

Our findings suggest that persistent Wnt pathway signalling results in activation of a potent negative feedback loop, involving upregulation of Wnt antagonists, that in physiological circumstances is likely to regulate Wnt levels. Of note, whereas Apc^-/- cells are insensitive to Wnt modulation at the receptor level,Apc-proficient cells are not, resulting in loss of stem cell features. To determine which antagonists are responsible for the observed effect, we treatedWT organoids with CM generated from cells overexpressing Notum, Wif1 or Dkk2, or with the recombinant variants (Figs. 3g, h, Extended Data Figs. 4a-f). We found that all three antagonists had the ability to reduce expansion and clonogenicity of WT organoids, with the most potent effect observed in combination. Analogously, co-culture of WT organoids with Apc^-/- organoids deficient in either Notum, Wif1 or Dkk2 (generated using CRISPR/Cas9) did not rescue WT organoid expansion (Extended Data Figs. 4g, h). In addition, CM derived from Wnt-antagonist depleted Apc^-/- organoids did not alleviate the reduction in Wnt signalling in our TOP-GFP reporter cell line (Extended Data Fig. 4i). However, titration of the CM from CRISPR/Cas9 KO Apc^-/- organoids lacking the three individual factors, established that the cultures deficient in Notum most rapidly lost the ability to reduce clonogenicity in WT organoids (Extended Data Fig. 4j). Together, these data indicate that none of the individual, upregulated Wnt antagonists is solely responsible for the observed inhibitory effects that Apc-deficient cells exert on their neighbours, but that Notum might be most critical in this context. The relevance of NOTUM was also confirmed in human organoids (Extended Data Fig. 4k, l). Given the central importance of the Wnt pathway in regulation of gut homeostasis, redundancy in molecules controlling the negative feedback is expected. This is in agreement with the accompanying manuscript by Flanagan et al. showing a marked increase in expression of other Wnt antagonists, including Wif1 and Dkk3, after loss of Notum in Apc^-/- organoids and adenomas. We reasoned that rendering WT cells insensitive to the Wnt antagonists by downstream Wnt activation, could provide an effective strategy to reduce the supercompetitor features of Apc- mutant cells. Indeed, organoids which were treated with GSK3β inhibitors LiCl or CHIR, or that expressed a constitutive active variant of β-catenin, were resistant to the Apc^-/- CM induced reduction in proliferation and clonogenicity (Fig. 3i, Extended Data Figs. 5a-d). Moreover, LiCl administration also rescued loss of stemness and clonogenicity in human colon organoids incubated with FAP CM (Extended Data Figs. 5e-f). This further supports the suggestion that boosting Wnt activation in wild type cells might be a promising approach to limit the competitive benefit of APC-mutant clones in humans.

LiCl rescues Wnt inhibition in vivo

Translation of these in vitro findings to an in vivo model was facilitated by the highly specific upregulation of Notum in Apc-deficient cells (Figs. 4a, b, Extended Data Figs. 6a-c). Analysis of sequential crypt bottom slices using Notum in situ hybridisation showed exclusive expression of Notum in homozygous recombined Apc (Exon14-Exon16) crypts, providing a direct read-out of bi-allelic (Notum^pos;E14/16^pos) Apc loss (Figs. 4a, b, Extended Data Fig. 6b). Importantly, the previously reported expression of Notum in (aged) Paneth cells did not impact our analyses due to markedly lower Notum levels in these cells as compared to Apc^-/- clones (Extended Data Fig. 6c)¹¹.

In the accompanying study by Flanagan et al. it is demonstrated that co-deletion of Notum together with Apc reduces the expansion rate of Apc-mutant clones, suggesting a direct involvement of Wnt antagonists in intestinal transformation. Given the observed functional redundancy in secreted Wnt ligands, we here evaluated if downstream pharmacological activation of the Wnt pathway in WT cells could limit the effects induced in the environment of Apc^+/- clones as well as their expansion within the crypt. To this end we employed a model system we previously developed to quantify the effect of oncogenic mutations on ISC dynamics in vivo³. First, we confirmed that oral LiCl treatment in Lgr5-Cre^Ert2; Rosa26^mTmG mice resulted in well-tolerated serum concentrations and effective Wnt activation in intestinal epithelial cells (Extended Data Figs. 7a-d). Moreover, we studied neutral ISC competition in WT mice in the presence or absence of LiCl treatment and observed that LiCl had no influence on fundamental ISC dynamics (Extended Data Figs. 7e-l). This indicates that LiCl exposed crypts continue to demonstrate neutral drift dynamics. Next, we employed Lgr5-Cre^Ert2; Apc^fl/fl mice and detected that while Notum^Pos/Apc^-/- clones reduced Lgr5 expression within the same crypt and in directly neighbouring crypts, LiCl treatment of mice prevented this (Figs. 4c, d, and Extended Data Figs. 8a-d). This both directly confirms the ability of Apc-deficient cells to induce differentiation in vivo, as well as the ability of LiCl to prevent this.

To study the impact of LiCl on the clonal dynamics of Apc-mutant clones, we again used Lgr5-Cre^Ert2; Apc^fl/fl mice and evaluated Notum^Pos/Apc^-/- clone size distributions within crypt bottoms at predefined days following tamoxifen injection, in the absence or presence of LiCl (Figs. 4e-g). Treatment with LiCl significantly reduced the rate of Notum^Pos/Apc^-/- clone expansion and fixation compared to the non-treated mice (Figs. 4h, i). In addition, LiCl reduced the probability of replacement (P_R) of WT ISCs by Apc^-/- ISCs from 0.65 (95% CI: 0.62-0.68) to 0.34 (95% CI: 0.31-0.37) (Extended Data Fig. 9a). This reduction is even below the initially expected return to neutral competition, corresponding to a P_R of 0.5. Further analyses indicated that this is caused by the fact that while Notum^Pos/Apc^-/- clones reduced the number of WT stem cells (N^WT) in crypts to an average of 4.7 (95% CI: 4.5-4.8), as compared to 5.6 (95% CI: 5.2-5.9) in fully WT crypts, in LiCl treated mice the number of functional ISCs is increased to 6.5 (95% CI: 6.2-6.8) (Extended Data Fig. 9b-d). Together, this results in a markedly reduced probability of clonal fixation in a crypt (P_fix) (Fig. 4j). This analysis was directly corroborated by a reduced number of Notum^Pos clones in LiCl treated mice (Fig. 4k). To evaluate if the observed effect of LiCl on mutant ISC dynamics is specific for Apc-deficient clones, and in light of the well-described competitive advantage of Kras^G12D-mutant cells^3,12, we analysed Kras^G12D-mutant clones in vitro and in vivo in the presence or absence of LiCl (Extended Data Fig. 10). We found Kras^G12D clone dynamics to remain unaffected by treatment with LiCl (Extended Data Figs. 10f-m), indicating that the reduction in competitive advantage of Apc-deficient clones is indeed related to antagonizing the specific effect of Apc-mutant clones.

Finally, we evaluated if the reduction in clonal fixation rate of Apc^-/- clones also resulted in a reduction of adenoma formation. To this end, we pre-treated Lgr5-Cre^Ert2; Apc^fl/fl mice with LiCl, induced low-level Apc-inactivation and maintained mice on LiCl treatment. Sixty days after induction we sacrificed the mice and evaluated the number of adenomas (Figs. 4l-n). This experiment revealed markedly reduced adenoma formation in all segments of the intestine (Fig. 4o, Extended Data Fig. 9e), and confirms the potency of rendering WT cells insensitive to the supercompetition effect of Apc-mutant clones in preventing intestinal tumour formation.

Discussion

In this study, we showed that Apc-mutant cells display supercompetitor properties as they actively drive elimination of WT ISCs from the crypt. To date, the best studied examples of supercompetitors are Minute-mutant and Myc-overexpressing cells in Drosophila, which both induce apoptosis in wild type cells^13–16. In addition, APC-mutant clones in the Drosophila midgut were demonstrated to actively induce apoptosis in the surrounding tissue¹⁷. Here, we detected that Apc-mutant cells induce differentiation of wild type ISCs through secretion of multiple Wnt antagonists. In agreement, supercompetition by means of differentiation has been described for Drosophila ovarian germline stem cells¹⁸, shown to be important for maintaining tissue integrity during murine skin homeostasis¹⁹, and is proposed to be the main mechanism of competition in adult tissues²⁰. Wnt antagonist Notum has also been determined to be the responsible driver of cell competition in the Drosophila wing imaginal disc²¹. Moreover, secretion of NOTUM by Paneth cells has recently been implicated in reducing stem cell function in the aging intestine, and pharmacological inhibition of NOTUM was shown to rejuvenate the intestine¹¹. Previously, many different Wnt antagonists have been reported to be upregulated in cells following genetic events leading to Wnt activation^22–26. In the present work, in conjunction with the study by Flanagan et al., we reveal their previously unrecognized contribution to intestinal tumour formation. Of note, we also confirm key aspects of the supercompetitor phenotype in a human context. In APC-deficient human cells we detected expression of an even larger set of partially redundant Wnt antagonists including NOTUM, DKK1, SFRP5 and WIF1 (Fig. 3h). This finding is in support of our approach to aim for pharmacological Wnt activation downstream of the ligand-receptor level, e.g. using LiCl. Timing of the administration of LiCl is critical as this strategy is only predicted to prevent tumour initiation and not growth, and therefore chemoprevention should be initiated at a young age²⁷. Interestingly, explorative epidemiological data of patients with bipolar disorder are in line with our conclusion that lithium has cancer preventive effects, specifically in digestive cancers^28,29. Our findings immediately provide a potentially potent novel strategy in reducing cancer incidence in individuals at high risk of developing intestinal cancers, in particular in patients with FAP that are characterized by germline APC mutations. More generally, identifying and counteracting signals from pre-malignant clones exerting a supercompetitor phenotype, might be a potent chemopreventive strategy in various cancer syndromes.

References

1. Fearnhead, N. S., Britton, M. P. & Bodmer, W. F. The ABC of APC. Human Molecular Genetics (2001) doi:10.1093/hmg/10.7.721.
2. Muzny, D. M. et al. Comprehensive molecular characterization of human colon and rectal cancer. Nature (2012) doi:10.1038/nature11252.
3. Vermeulen, L. et al. Defining stem cell dynamics in models of intestinal tumor initiation. Science 342, 995–8 (2013).
4. Fearon, E. F. & Vogelstein, B. for Colorectal Tumorigenesis. Cell 61, 759–767 (1990).
5. Morin, P. J., Vogelstein, B. & Kinzler, K. W. Apoptosis and APC in colorectal tumorigenesis. Proc. Natl. Acad. Sci. (2002) doi:10.1073/pnas.93.15.7950.
6. Fevr, T., Robine, S., Louvard, D. & Huelsken, J. Wnt/beta-catenin is essential for intestinal homeostasis and maintenance of intestinal stem cells. Mol. Cell. Biol. (2007) doi:10.1128/MCB.01034-07.
7. Boone, P. G. et al. A cancer rainbow mouse for visualizing the functional genomics of oncogenic clonal expansion. Nat. Commun. 10, 5490 (2019).
8. Albuquerque, C. et al. The ‘just-right’ signaling model: APC somatic mutations are selected based on a specific level of activation of the beta-catenin signaling cascade. Hum. Mol. Genet. (2002) doi:10.1093/hmg/11.13.1549.
9. Crabtree, M. et al. Refining the relation between ‘first hits’ and ‘second hits’ at the APC locus: The ‘loose fit’ model and evidence for differences in somatic mutation spectra among patients. Oncogene (2003) doi:10.1038/sj.onc.1206471.
10. Haber, A. L. et al. A single-cell survey of the small intestinal epithelium. Nature (2017) doi:10.1038/nature24489.
11. Pentinmikko, N. et al. Notum produced by Paneth cells attenuates regeneration of aged intestinal epithelium. Nature (2019) doi:10.1038/s41586-019-1383-0.
12. Snippert, H. J., Schepers, A. G., Van Es, J. H., Simons, B. D. & Clevers, H. Biased competition between Lgr5 intestinal stem cells driven by oncogenic mutation induces clonal expansion. EMBO Rep. 15, 62–69 (2014).
13. Morata, G. & Ripoll, P. Minutes: Mutants of Drosophila autonomously affecting cell division rate. Dev. Biol. (1975) doi:10.1016/0012-1606(75)90330-9.
14. de la Cova, C. et al. Drosophila Myc Regulates Organ Size by Inducing Cell Competition. 117, 107–116 (2004).
15. Moreno, E., Basler, K. & Morata, G. Cells compete for decapentaplegic survival factor to prevent apoptosis in Drosophila wing development. Nature (2002) doi:10.1038/416755a.
16. Moreno, E. & Basler, K. dMyc transforms cells into super-competitors. Cell (2004) doi:10.1016/S0092-8674(04)00262-4.
17. Suijkerbuijk, S. J. E., Kolahgar, G., Kucinski, I. & Piddini, E. Cell Competition Drives the Growth of Intestinal Adenomas in Drosophila. Curr. Biol. 26, 428–438 (2016).
18. Jin, Z. et al. Differentiation-Defective Stem Cells Outcompete Normal Stem Cells for Niche Occupancy in the Drosophila Ovary. Cell Stem Cell (2008) doi:10.1016/j.stem.2007.10.021.
19. Liu, N. et al. Stem cell competition orchestrates skin homeostasis and ageing. Nature (2019) doi:10.1038/s41586-019-1085-7.
20. Ellis, S. J. et al. Distinct modes of cell competition shape mammalian tissue morphogenesis. Nature (2019) doi:10.1038/s41586-019-1199-y.
21. Vincent, J. P., Kolahgar, G., Gagliardi, M. & Piddini, E. Steep Differences in Wingless Signaling Trigger Myc-Independent Competitive Cell Interactions. Dev. Cell (2011) doi:10.1016/j.devcel.2011.06.021.
22. Kakugawa, S. et al. Notum deacylates Wnt proteins to suppress signalling activity. Nature (2015) doi:10.1038/nature14259.
23. Koo, B. K. et al. Tumour suppressor RNF43 is a stem-cell E3 ligase that induces endocytosis of Wnt receptors. Nature (2012) doi:10.1038/nature11308.
24. De Robertis, M. et al. Novel insights into Notum and glypicans regulation in colorectal cancer. Oncotarget (2015) doi:10.18632/oncotarget.5652.
25. González-Sancho, J. M. et al. The Wnt antagonist DICKKOPF-1 gene is a downstream target of β-catenin/TCF and is downregulated in human colon cancer. Oncogene (2005) doi:10.1038/sj.onc.1208303.
26. Niida, A. et al. DKK1, a negative regulator of Wnt signaling, is a target of the β-catenin/TCF pathway. Oncogene (2004) doi:10.1038/sj.onc.1207892.
27. Gould, T. D., Gray, N. A. & Manji, H. K. Effects of a glycogen synthase kinase-3 inhibitor, lithium, in adenomatous polyposis coli mutant mice. Pharmacol. Res. (2003) doi:10.1016/S1043-6618(03)00096-3.
28. Martinsson, L., Westman, J., Hällgren, J., Ösby, U. & Backlund, L. Lithium treatment and cancer incidence in bipolar disorder. Bipolar Disord. (2016) doi:10.1111/bdi.12361.
29. Huang, R.-Y., Hsieh, K.-P., Huang, W.-W. & Yang, Y.-H. Use of lithium and cancer risk in patients with bipolar disorder: population-based cohort study. Br. J. Psychiatry 209, 393–399 (2016).

Materials and Methods

Animal experiments

Lgr5-EGFP-IRES-Cre^ERT2, Villin-Cre^ERT2, Rosa26mTmG, Apc^fl/fl and Kras^G12D mice have been described earlier^30–34. All in vivo experiments were approved by the animal experimentation committee at the Amsterdam UMC - location Academic Medical Center in Amsterdam under nationally registered licence number AVD1180020172125 and performed according to national guidelines. Mice were housed in a 12 hour light/ 12 hour dark cycle, with temperatures between 20-24°C and 40-70% humidity. For short term assays, both male and female mice were used. For long term assays, only females were used to prevent the risk of preliminary dropout due to fighting male mice. All mice were between 6-12 weeks old at the start of the experiments. For all mouse experiments, sufficient sample sizes were determined based on previous studies with a similar study design^3,35. Experimental animals were randomly assigned to the control or lithium treated groups, clone sizes and adenoma counts were scored blindly. In vivo low-dose recombination was induced by intraperitoneal injection (i.p.) of 0.3 mg (for Rosa26^mTmG and Kras^G12D mice) or 2mg (for Apc^fl/fl mice) tamoxifen (Sigma) dissolved in sunflower oil. Mice were either assigned to a control group or treated with lithium chloride (LiCl, Sigma) dissolved at a final concentration of 300 mg/liter in tap water. Treatment with LiCl was initiated 7 days before recombination by i.p. injection of tamoxifen and was administered until the day they were sacrificed. For short term experiments to study stem cell dynamics, mice were sacrificed at day 4, 7, 10, 14, 21 days after intra peritoneal (i.p.) injection and intestines were removed and further processed for analyses. For long term adenoma formation experiments, mice were injected with 2 mg tamoxifen and sacrificed 60 days after i.p. injection. Mouse discomfort during tumour formation assays was closely monitored, and endpoints were determined as < 15% weight loss within 2 days or a mouse grimace scale (MGS) score < 3. These endpoints were not exceeded during this study. After 60 days, intestines were removed and polyps were counted macroscopically.

Tissue processing & clone size quantification

After the mice were sacrificed, intestines were removed fully and washed thoroughly with ice-cold PBS. The intestines were cut into pieces of 5 mm, opened longitudinally and fixed overnight in 4% paraformaldehyde (PFA) solution. To preserve tissue integrity the intestines were kept in 30% sucrose solution for another night before freezing. Crypt bottoms were sliced with a thickness of 10mm at a Cryostar™ NX70 cryostat and placed on glass slips and counterstained with Hoechst-33342. Fluorescent lineage tracing labels were visualized with a SP8X Confocal (Leica) using Leica Application Suite (LAS) software. RNA-ISH stained coupes were counterstained with hematoxylin and scanned using the IntelliSite Ultra Fast 1.6 slide scanner (Philips). For all crypt analyses, clone sizes were quantified as proportions of the crypt circumference (in eights, 1:8-8:8).

Organoid culture

Murine intestinal crypts were isolated from Lgr5-EGFP-IRES-Cre^ERT2, Lgr5-EGFP-IRES-Cre^ERT2;Apc^fl/fl or Villin-Cre^ERT2;Apc^fl/fl mice as described by Sato et al.³⁶. In short, intestines were removed from the mouse and washed thoroughly with ice-cold PBS. Next, the intestine was opened longitudinally and the villi were gently scraped off by a glass cover slide. The intestine was cut into pieces of 5x5 mm and incubated in 2 mM EDTA solution for 30 min. at 4°C. After removal of the EDTA, the crypts were resuspended in ice-cold 1% FCS in PBS by vigorously shaking the tube and passing the supernatant through a 70mm strainer. Isolated crypts were resuspended in MatrigelÒ (Corning) and seeded in pre-heated 24-well plates, supplemented with basal organoid medium consisting of advanced DMEM/F12 medium (Gibco) containing 100X N2 and 50X B27 supplements, 100X Glutamax, 5 mM HEPES, 1 mM N-acetyl-L-cysteine (Sigma), and 100X antibiotic/antimycotic (all Gibco). The basal organoid medium was freshly supplemented with the following growth factors: mouse EGF 50 ng/ml (TEBU-BIO), R-spondin (conditioned medium), Noggin (conditioned medium). The first two days after crypt isolation CHIR99021 (Axon Medchem) and ROCK inhibitor (Sigma) was added to the medium. Ctnnb1^S organoids expressing a constitutive active variant of b-catenin were generated as described by Adam et al.³⁷. For in vitro recombination of loxP flanked alleles, 1mM 4OH-Tamoxifen (Sigma) was added to the medium. Recombination of the Apc gene was validated by digital droplet PCR (Bio-Rad) using EvaGreenâ Supermix (Bio-Rad). Lgr5-EGFP-IRES-Cre^ERT2 organoids, referred to as wild type (WT) organoids, were stably transduced with a red fluorescent mCherry construct (LeGO-C2, Addgene #27399). The in vitro recombined Lgr5-EGFP-IRES-Cre^ERT2;Apc^fl/fl, referred to as Apc^-/-, were stably transduced with a green fluorescent Venus construct (LeGO-V2, Addgene #27340). During competition assays, organoids were plated in equal numbers in 24-well plates and full wells were scanned over time by the EVOS FL Cell Imaging System (Thermo Scientific). Conditioned medium (CM) was taken from 2-3 day old WT or Apc^-/- organoid cultures and freshly supplemented with growth factors R-spondin, Noggin and mEGF. During the competition assays the medium (normal and conditioned) was replaced every other day to minimize effects of medium depletion. To assess the effect of medium depletion, CM was 10x concentrated using 10 kDa Amicon® centrifugal filters (Millipore) and added to fresh ENR medium as 1:10. GSK3b inhibition in the CM transfer assays was performed by administering 5mM LiCl or 2.5mM CHIR99021 to the medium. Recombinant murine proteins NOTUM (2 mg/mL, R&D, 9150-NO-050), WIF1 (5 mg/mL, R&D, 135-WF-050) and DKK2 (1 mg/mL, R&D, 2435-DKB-010), and human recombinant NOTUM( 0.1-1 mg/mL, R&D, 9118-NM-050) were freshly added to the culture medium, medium was refreshed every other day.

Human organoid cultures were derived from normal colonic tissue obtained from resection material and from polyps of patients diagnosed with familial adenomatous polyposis (FAP)³⁸. The collection of normal and adenomatous material from the colon was approved by the Medical Ethical Committee of Academic Medical Center (AMC), under approval numbers 2014/178 (normal tissue) and MEC 09/146 (adenoma tissue). Tissue was collected following written informed consent of patients, and the experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report. Approval for the use of this material has been given for the previous and current study. Normal colon organoids were isolated and processed as previously described⁶. FAP organoids were generated by cutting the polyps into small pieces and plating them into Matrigel (Corning) and were described previously³⁸. Both normal and FAP organoids were cultured in basal organoid medium as described above, freshly supplemented with 10 mM Nicotinamide (Sigma), 10 mg/mL gentamicin (Lonza), 3 mM SB202190 (Sigma), 500 nM A83-01 (Tocris), 10 nM Prostaglandin E2 (Santa Cruz Biotechnology), 10 nM Gastrin (Sigma), 20 ng/mL human EGF (Peptrotech), R-spondin and Noggin. Normal colon organoids were additionally supplemented with Wnt3a (conditioned medium). Medium was refreshed every 2 days.  

CRISPR cloning

To generate CRISPR KO lines for Notum, Wif1 and Dkk2, two different sgRNA’s were designed for each gene using Benchling and sequences can be found in Supplementary Table 1. The sgRNA oligos were cloned into the lentiCRISPR v2 plasmid (Addgene #52961) and transformed using Stabl3 competent bacteria (Invitrogen). Successful cloning of the guides was verified using Sanger sequencing. Lentiviral particles were generated using third generation packaging plasmids pMDLg/pRRE (Addgene #12251), pRSV-Rev (Addgene #12253) and MD2.G (Addgene #12259). Organoids were transduced with plasmids containing viral particles containing two sgRNA’s for one gene, to accommodate the disruption of the target gene through large editing events. After puromycin selection, organoids were single cell sorted to generate unique KO clones, that were validated for editing by Sanger sequencing and TIDE analysis³⁹.

Cell Culture

Mouse embryonic fibroblasts (MEFs, ATCC) and HEK293T (ATCC) cells were both cultured in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal calf serum (FCS), 1% Glutamine, and antibiotic penicillin and streptomycin. Cells were maintained at 37°C in humidified air containing 5% CO₂. All cell lines were routinely checked for mycoplasm contamination, no cell line authentication was performed.

Generation of overexpression constructs

To generate overexpression lines for Notum, Wif1 and Dkk2 RNA was isolated from Apc-mutant intestinal organoids. cDNA was generated using SuperScript III RT (Sigma) and ORFs for Notum, Wif1 and Dkk2 were PCR amplified using primers containing EcorI and NotI restriction digestion sites. Primer sequences can be found in Supplementary Table 2. Amplified ORFs were cloned into lentiviral plasmid LegO-V2 (Addgene #27340). Lentiviral particles were generated as described above. The viral particles were transduced into HEK293T cells and Venus-positive cells were selected by FACS sorting. Expression levels of Notum, Wif1 and Dkk2 were assessed using RT-qPCR and protein levels were validated using ELISA for NOTUM (LS-F17999) and WIF1 (LS-F39936-1, LS Biosciences).

TOP-GFP assay

MEFs were stably transduced with Wnt reporter TOP-GFP (Addgene plasmid # 35491). For TOP-GFP assays, cells were stimulated with Wnt3A conditioned medium for 24 hours after which GFP positivity was measured by flow cytometry. For downstream Wnt activation, either 5 mM LiCl or 2.5 mM CHIR99021 was supplemented to the medium.

RNA isolation & qPCR

RNA was extracted using the Bioke Nucleospin RNA isolation kit (cat no. 740955). Complementary DNA syntheses was generated using SuperScript III RT (Sigma). Sybr Green (Roche) RT-qPCR reactions were performed with the Roche LightCycler 480 system under standard conditions. The DDCt method was used to calculate gene expression. All DDCt values were normalized to housekeeping genes Rpl37 and Hprt. Primer sequences can be found in Supplementary Table 3. 

Western blotting

Organoids were harvested in Cell Recovery solution (Corning) and incubated on ice for 30 min to remove Matrigel remnants. Following 2 PBS washes, protein lysates were made using 10X cell lysis buffer (Cell Signaling Technologies) according to manufactureres’ protocol. Protein concentrations were determined using Pierceä Protein Assay Kit (Thermo Scientific), and 30 μg protein was loaded in 4-15% Mini-PROTEANâ TGX precast protein gels (Bio-Rad), separated by electrophoresis and transferred to PDVF membranes using the Trans-Blot Turbo System (Bio-Rad). Next, membranes were blocked in 5% Skim Milk Powder (Sigma) before they were incubated with primary antibodies in 5% BSA/TBST overnight at 4°C on a roller bank. The next day, the membranes were incubated with HRP-conjugated secondary antibodies for 1h at room temperature in 5% BSA/TBST. Protein levels were detected using Pierceä ECL Western Blotting Substrate (Thermo Scientific) and revealed using ImageQuant LAS 4000 (GE Healthcare Life Sciences). Primary antibodies used are anti-β-Catenin (9562, Cell Signaling Technologies, 1:1000) and anti-GAPDH (MAB374, Millipore, 1:1000). Secondary antibodies are anti-rabbit-HRP (7074, Cell Signaling technologies) and anti-mouse-HRP (1070-05, Southern Biotech).

Flow Cytometry

All FACS analysis experiments were performed on the BD LSRFortessaä (BD Biosciences). FACS sorting was performed on the BD FACSAriaä III Cell Sorter (BD Biosciences). In vitro Lgr5-GFP^high populations were gated on Dapi^neg population. In vivo Lgr5-GFP^high populations were gated on Hoechst^neg, EPCAM^pos (anti-mouse CD326 (EPCAM)-APC (17-5791-82, Bioscience, 1:100)) populations. Absolute cell numbers were determined using BD Trucount tubes (BD Biosciences). Data acquisition was performed using FACSDiva software V8 (BD Biosciences), data analysis was performed using FlowJo software (Flowjo, LLC). FACS gating strategies can be found in Supplementary Figure 2.

Immunofluorescence

Murine stainings were performed on fixed frozen and paraffinized tissues. Human stainings were performed on paraffined biopsies derived from FAP patients, all biopsies were scored by a pathologist for adenomatous lesions. Prior to staining, paraffin coupes were deparaffinized and treated with antigen retrieval in citrate solution (pH 6.0). Next, samples were blocked using ultra-V blocking solution (Immunologic). Primary antibodies were administered in antibody diluent (ScyTek) and incubated overnight at 4°C. Slides were washed thoroughly and incubated in secondary antibody for 1h at room temperature. Hoechst-33342 (Thermo Scientific) was used as nuclear counterstain and was incubated at 10mg/ml for 5 min at room temperature before slides were covered with Prolong™ Gold antifade reagent (Invitrogen) and sealed with coverslips (VWR). All stainings were analysed using the SPX8 Confocal (Leica) and stored at 4°C. The following antibodies were used: anti-mouse MUC2 (sc-15334, Santa Cruz, 1:100), anti-human E-Cadherin (AF748, R&D Systems, 1:200), anti-rabbit Alexa Fluor 488 (A11034, Invitrogen, 1:500), and anti-goat Alexa Fluor 488 (51475A, Invitrogen, 1:500).

RNA in situ hybridisation

RNA in situ hybridisation (RNAscope) and Basescope was performed on fixed frozen intestinal tissue (mouse) and paraffine embedded tissue (mouse and human) according to manufacturer’s protocol (ACD RNAscope 2.5 HD – Brown and Red, and ACD BaseScopeÔ v2 - Red). RNAscope was used for detection of mouse Notum (#428981), Wif1 (#412361), Dkk2 (#404841) or positive control Ppib (#313911) and human NOTUM (#430311) and positive control PPIB (#313901). BaseScope probe ApcE14-E16 (#703011) was used to detect recombined Apc alleles. RNAscope duplex was performed using the RNAscope Duplex Reagent Kit (#322430, ACD) with additional Lgr5 probe (#312171-C2). After RNAscope procedures tissues were counterstained for Hematoxylin or Hoechst-33342. RNAscope was quantified using QuPath software v0.2.2⁴⁰.

RNA-Seq

Organoid RNA sequencing libraries were prepared using the KAPA RNA Hyperprep with RiboErase (Roche) following manufacturer’s protocol. Total RNA isolation was performed by trizol-chloroform extraction in combination with the RNeasy MinElute Cleanup Kit (Qiagen, Hilden, Germany). RNA integrity was assessed with the Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA). Libraries were barcoded, quantified using NEBNext® Library Quant Kit for Illumina (New England Biolabs (NEB), MA, USA), pooled equimolarly and multiplex sequenced (single-end 50 bp reads) on the Illumina Hiseq4000 platform.

RNA-seq data analysis

Sequence read quality was assessed by means of the FastQC method (v0.11.5; http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Trimmomatic version 0.36 was used to trim Illumina adapters and poor-quality bases (trimmomatic parameters: leading=3, trailing=3, sliding window=4:15, minimum length=40)⁴¹. The remaining high-quality reads were used to align against the Genome Reference Consortium mouse genome build 38 (GRCm38)⁴². Mapping was performed by HISAT2 version 2.1.0 with parameters as default⁴³. Count data were generated by means of the HTSeq method, and analysed using the DESeq2 method in the R statistical computing environment (R Core Team 2014. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria)^44,45. Statistically significant differences were defined by Benjamini & Hochberg adjusted probabilities < 0.05.

Data visualization

Heatmaps and Volcano plots from publicly available datasets GSE145308⁴⁶, GSE65461⁴⁷ and GSE8671⁴⁸ were analysed using Genomics Analysis and Visualization Platform R2⁴⁹. Signatures scores were based on published gene signatures for Wnt signalling “HALLMARK_WNT_BETA_CATENIN_SIGNALING” (M5895), van der Flier⁵⁰ and intestinal stem cells signatures by Muñoz⁵¹ and Merlos-Suárez⁵².

Stem cell drift modelling

The clone data was modelled using the models and methods developed in Vermeulen et al.³. An R package implementing the model was used and is available at  https://github.com/MorrisseyLab/CryptDriftR. Briefly, the clonal dynamics generated by the stem cells are modelled as a one-dimensional discrete random walk with absorbing states at 0 and N, where N is the total number of stem cells^3,53. When modelling mutant stem cells, the balance between replacing neighbours or being replaced is inferred directly from the data. The fitting of the stochastic model to the data is done using a Bayesian approach with a multinomial likelihood for the counts of the different clone sizes measured.

In light of our experimental observations on stem cell numbers we adapt the model to include the change in WT stem cell numbers as a mode of mutant advantage. We utilise the same base drift model, however to add granularity we consider the change in stem cell numbers to be distributional rather than a single change. This is done by modelling the full distribution as a mixture of drift models with different numbers of stem cells, where the mixing weights are to be inferred:

Q(λ, τ, t)= ∑_n=3^Nmax α_n P_n(λ, τ, t)

In order to constrain the degrees of freedom of the model we link the  by modelling as a gaussian with a mean and a variance to be inferred and later normalising so that . The model is implemented in Stan and distributions are produced using HMC⁵⁴.

Statistics and reproducibility

All in vitro organoid monocultures were quantified blindly using ImageJ FIJI v2.0⁵⁵. This was impossible for the co-culture experiments due to the fluorescent features of these co-cultures. All in vivo histological data was scored blindly. Visualization and statistical analysis of data was performed using Graphpad Prism, where most data was analysed using two-sided Student’s t-test. In case other statistical tests were applied this was noted in the figure legends. All RNAscopes were performed on at least 3 independent biological samples (either 3 different mice or patients).

Data availability statement

The sequence libraries generated in this study are publicly available through the National Center for Biotechnology Information (NCBI) Gene Expression Omnibus (GEO) under accession GSE144325: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144325. Other datasets used in this study are also publicly available via NCBI GEO under accession numbers GSE145308, GSE65461, and GSE8671.

Code availability statement

The clone data was modelled using an R package implementing the model and is available at  https://github.com/MorrisseyLab/CryptDriftR.

Method references

30. Barker, N. et al. Identification of stem cells in small intestine and colon by marker gene Lgr5. Nature 449, 1003–1007 (2007).
31. El Marjou, F. et al. Tissue-specific and inducible Cre-mediated recombination in the gut epithelium. Genesis (2004) doi:10.1002/gene.20042.
32. Shibata, H. et al. Rapid colorectal adenoma formation initiated by conditional targeting of the APC gene. Science (80-. ). 278, 120–133 (1997).
33. Muzumdar, M. D., Tasic, B., Miyamichi, K., Li, N. & Luo, L. A global double-fluorescent cre reporter mouse. Genesis (2007) doi:10.1002/dvg.20335.
34. Johnson, L. et al. Somatic activation of the K-ras oncogene causes early onset lung cancer in mice. Nature (2001) doi:10.1038/35074129.
35. Van Der Heijden, M. et al. Bcl-2 is a critical mediator of intestinal transformation. Nat. Commun. 7, 1–11 (2016).
36. Sato, T. et al. Long-term expansion of epithelial organoids from human colon, adenoma, adenocarcinoma, and Barrett’s epithelium. Gastroenterology (2011) doi:10.1053/j.gastro.2011.07.050.
37. Adam, R. S. et al. Intestinal region-specific Wnt signalling profiles reveal interrelation between cell identity and oncogenic pathway activity in cancer development. Cancer Cell Int. (2020) doi:10.1186/s12935-020-01661-6.
38. Fessler, E. et al. TGFβ signaling directs serrated adenomas to the mesenchymal colorectal cancer subtype. EMBO Mol. Med. (2016) doi:10.15252/emmm.201606184.
39. Brinkman, E. K., Chen, T., Amendola, M. & Van Steensel, B. Easy quantitative assessment of genome editing by sequence trace decomposition. Nucleic Acids Res. (2014) doi:10.1093/nar/gku936.
40. Bankhead, P. et al. QuPath: Open source software for digital pathology image analysis. Sci. Rep. (2017) doi:10.1038/s41598-017-17204-5.
41. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics (2014) doi:10.1093/bioinformatics/btu170.
42. Harrow, J. et al. GENCODE: producing a reference annotation for ENCODE. Genome Biol. (2006).
43. Kim, D., Langmead, B. & Salzberg, S. L. HISAT: A fast spliced aligner with low memory requirements. Nat. Methods (2015) doi:10.1038/nmeth.3317.
44. Anders, S., Pyl, P. T. & Huber, W. HTSeq-A Python framework to work with high-throughput sequencing data. Bioinformatics (2015) doi:10.1093/bioinformatics/btu638.
45. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. (2014) doi:10.1186/s13059-014-0550-8.
46. Ringel, T. et al. Genome-Scale CRISPR Screening in Human Intestinal Organoids Identifies Drivers of TGF-β Resistance. Cell Stem Cell (2020) doi:10.1016/j.stem.2020.02.007.
47. Reed, K. R. et al. Hunk/Mak-v is a negative regulator of intestinal cell proliferation. BMC Cancer (2015) doi:10.1186/s12885-015-1087-2.
48. Sabates-Bellver, J. et al. Transcriptome profile of human colorectal adenomas. Mol. Cancer Res. (2007) doi:10.1158/1541-7786.MCR-07-0267.
49. R2 database. Genomic Analysis and Visualization Platform. The Department of Human Genetics in the Amsterdam Medical Centre (AMC) (2019).
50. Van der Flier, L. G. et al. The Intestinal Wnt/TCF Signature. Gastroenterology (2007) doi:10.1053/j.gastro.2006.08.039.
51. Muñoz, J. et al. The Lgr5 intestinal stem cell signature: Robust expression of proposed quiescent ′ +4′ cell markers. EMBO J. (2012) doi:10.1038/emboj.2012.166.
52. Merlos-Suárez, A. et al. The intestinal stem cell signature identifies colorectal cancer stem cells and predicts disease relapse. Cell Stem Cell 8, 511–524 (2011).
53. Lopez-Garcia, C., Klein, A. M., Simons, B. D. & Winton, D. J. Intestinal stem cell replacement follows a pattern of neutral drift. Science 330, 822–825 (2010).
54. Carpenter, B. et al. Stan: A probabilistic programming language. J. Stat. Softw. (2017) doi:10.18637/jss.v076.i01.
55. Schindelin, J. et al. Fiji: An open-source platform for biological-image analysis. Nature Methods vol. 9 676–682 (2012).

Acknowledgements

S.M.v.N. is supported by a NWO OOA PhD scholarschip (022.005.002). This work is supported by The New York Stem Cell Foundation and grants from KWF (UVA2014-7245), the Maurits en Anna de Kock Stichting (2015-2), Worldwide Cancer Research (14-1164), the Maag Lever Darm Stichting (MLDS-CDG 14-03), the European Research Council (ERG-StG 638193) and ZonMw (Vidi 016.156.308) to L.V. L.V. is a New York Stem Cell Foundation – Robertson Investigator. We would like to thank the AMC laboratory for clinical chemistry (LAKC), the mouse breeding and research facilities, and the core facilitiies for genomics, cellular imaging and pathology for their technical support.

Author Contributions

S.M.v.N. and L.V conceptualized the project. S.M.v.N. and L.V designed the experiments. S.M.v.N, N.E.G., L.E.N., M.S.v.D. D.R.S., performed in vitro organoid (co)culture experiments, DNA/RNA/protein assays, stainings and RNA-ISH. V.K. performed in vitro organoid recombination assays. F.V.B., assisted with FACS assays. P.R., and A.S.A., assisted with human organoid cultures and FAP adenoma data. M.F.v.B., generated the fluorescent organoid cultures. N.L. designed and generated overexpression constructs. N.E.G., L.E.N., M.S.v.D., and M.C.L. performed in vivo experiments and tissue processing. D.O.W. developed the CRISPR strategies. L.F.M., and S.t.H., generated clone size plots. P.M.K. generated spider plots and helped with visualization of the data. L.K. and E.D. provided human materials. B.P.S. and J.K. analysed RNA-seq data. E.M. designed and performed the mathematical modelling. S.M.v.N. and L.V. wrote the manuscript, with help from B.P.S., J.K., E.M., and N.L. J.P.M., D.J.W. and M.F.B. advised on the project.  

Competing interests

L.V. received consultancy fees from Bayer, MSD, Genentech, Servier and Pierre Fabre but these had no relation with the content of this publication.

Additional information

Correspondence and requests for materials should be addressed to L. Vermeulen: l.vermeulen@amsterdamumc.nl

For information regarding stem cell drift modelling please contact E. Morrissey: edward.morrissey@imm.ox.ac.uk

Reprints and permissions

No permissions are required to publish any images or illustrations.