Author response: Targeting mutant RAS in patient-derived colorectal cancer organoids by combinatorial drug screening

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Colorectal cancer (CRC) organoids can be derived from almost all CRC patients and therefore capture the genetic diversity of this disease. We assembled a panel of CRC organoids carrying either wild-type or mutant RAS, as well as normal organoids and tumor organoids with a CRISPR-introduced oncogenic KRAS mutation. Using this panel, we evaluated RAS pathway inhibitors and drug combinations that are currently in clinical trial for RAS mutant cancers. Presence of mutant RAS correlated strongly with resistance to these targeted therapies. This was observed in tumorigenic as well as in normal organoids. Moreover, dual inhibition of the EGFR-MEK-ERK pathway in RAS mutant organoids induced a transient cell-cycle arrest rather than cell death. In vivo drug response of xenotransplanted RAS mutant organoids confirmed this growth arrest upon pan-HER/MEK combination therapy. Altogether, our studies demonstrate the potential of patient-derived CRC organoid libraries in evaluating inhibitors and drug combinations in a preclinical setting. https://doi.org/10.7554/eLife.18489.001 eLife digest Recent technical advances mean that miniature replicas of many tissues can be grown in the laboratory. These so-called organoids provide scientists with model systems that are not as limited as simple, two-dimensional sheets of cells growing in a petri dish, and less labor and resource intensive than studies using laboratory animals. In particular, organoids grown from tumor cells from cancer patients have been suggested as having numerous advantages over both laboratory-grown cancer cells and mice when it comes to testing potential new anticancer drugs. Mutations in a gene called KRAS are common in many types of cancer including colon cancer. Tumors with these mutations are difficult to treat and so far virtually all attempts to generate compounds that selectively interfere with the KRAS protein encoded by the mutant gene have failed. Instead, drugs that indirectly inhibit this protein's effects by targeting other proteins in the same signaling pathway are currently being tested on patients. However, there is still a need for better ways to pre-test whether these drugs will be effective in humans without having to expose the patient to side effects or an ineffective drug. Now, Verissimo, Overmeer, Ponsioen et al. have tested clinically-used KRAS pathway inhibitors and drug combinations against normal colon organoids and colon cancer organoids derived from patients with colon cancer. Gene editing techniques were used to introduce KRAS mutations into some of the normal organoids grown from healthy tissue, and into cancer organoids grown from tumors that had a normal copy of the KRAS gene. In all cases, only those organoids with mutant forms of the KRAS gene were resistant to the treatments. Furthermore, when organoids with the KRAS mutation were treated with some combination therapies that are currently being tested in clinical trials, the tumors stopped growing but the tumor cells failed to die. Similar drug treatments on mice carrying human colon cancer organoids confirmed these results, which is in line with previous studies where tumor tissue from human patients was transplanted into mice. These findings show that collections of tumor organoids from multiple patients could help researchers to quickly identify and optimize targeted anticancer therapies before they are incorporated into clinical trials. In the future, clinical studies are needed to verify how accurately the testing of cancer drugs on organoids predicts whether the drug will or will not work in patients. https://doi.org/10.7554/eLife.18489.002 Introduction One of the great challenges in targeted cancer treatment has been the development of effective RAS-targeting drugs. RAS mutations occur in about 15% of all human tumors (Bos, 1989) and so far all attempts to selectively interfere in mutant RAS signaling have failed in the clinic (Stephen et al., 2014; Cox et al., 2014). Progress has long been impeded by the fact that the currently used model systems to pre-test drugs are insufficient: cell lines, on the one hand, have very limited genetic diversity, while mouse models on the other hand, may not represent human tumors (Sachs and Clevers, 2014; Gould et al., 2015). Moreover, until recently, personalized medicine required large-scale in-vitro screening on short-term cultures of tumor sections (Centenera et al., 2013), or alternatively, resource-intensive in-vivo screens using xenotransplantation of tumors into immunodeficient mice (Jin et al., 2010; Tentler et al., 2012). Recently, stem-cell based organoid technology was introduced to establish long-term cultures of both normal and tumor tissues from various organs (Sato et al., 2009, 2011; Bartfeld et al., 2015; Boj et al., 2015; Huch et al., 2015; Karthaus et al., 2014; Gao et al., 2014). The advantage of this technology is that it can capture the genetic diversity of both normal and tumor tissues. Indeed, for colorectal cancer (CRC) a genetically diverse Biobank of patient-derived CRC organoids was established and used to integrate genomic data and monotherapy drug responses at the level of individual patient-derived organoid lines (van de Wetering et al., 2015). We employed this biobank to further explore potential strategies to target mutant RAS, including the combination therapy of pan-HER and MEK inhibition, which is currently tested in clinical trials. We confirm the strong correlation between the presence of mutant RAS and resistance towards EGFR inhibition. Our data reinforce the notion that an oncogenic mutation in RAS is sufficient to confer this resistance independent of cellular status, whether it concerns normal or tumorigenic cells. Moreover, real-time imaging of the resistant drug response at the cellular level reveals predominant cell-cycle arrest in RAS mutant organoids, in contrast with the complete induction of cell death in CRC organoids with WT RAS. In vivo drug response of xenotransplanted RAS mutant CRC organoids confirmed the arrest in tumor growth upon dual inhibition of the EGFR-MEK-ERK pathway. Finally, efficient inhibition by dual targeting of the mutant RAS pathway strongly sensitizes for the induction of cell death, as illustrated by minimal addition of BCL inhibition. Our studies demonstrate the strong potential of patient-derived CRC organoid libraries in evaluating inhibitors and drug combinations in a preclinical setting. Results Drug response of patient-derived CRC organoids with and without mutant KRAS To explore drug responses of patient-derived CRC organoids towards combination therapies of targeted inhibitors of the EGFR-RAS-ERK pathway, we applied a drug sensitivity screen using EGFR-family and MEK inhibitors (EGFRi and MEKi resp.) either as mono or combination therapy on two cancer organoids from a previously established biobank of CRC organoids (van de Wetering et al., 2015). To start, we chose cancer organoids from the individuals P8 and P26, which share a similar composition of frequent cancer mutations such as functionally inactive APC and TP53. However, they differ in their KRAS status. P8T contains wild-type (WT) KRAS, while P26T contains an oncogenic mutant version of KRAS (G12V). 3D-organoids were challenged with drugs for 72 hr and drug responses were determined by quantifying cell viability through measurements of ATP levels using CellTiter-Glo (van de Wetering et al., 2015). We observed the expected sensitivity of P8T towards afatinib (irreversible EGFR/HER2 inhibitor) and insensitivity of KRAS mutant P26T (Figure 1A). Selumetinib (MEKi) as a monotherapy showed little efficacy in both P8T and P26T, but combination therapy confirmed previous findings that MEKi sensitizes RAS mutant tumor cells to EGFR/HER2 inhibition (Figure 1A) (Sun et al., 2014). However, the KRAS mutant P26T organoids were less sensitive to the combination therapy than the KRAS WT P8T organoids. Figure 1 with 1 supplement see all Download asset Open asset Drug responses of patient-derived CRC organoids with and without mutant KRAS. (A) Dose-response curves of patient-derived CRC organoids P8T (KRASWT; APC and TP53 mutant) and P26T (KRASG12V; APC and TP53 mutant) treated with the dual EGFR/HER2 inhibitor afatinib, MEK inhibitor selumetinib or a combination thereof. Cell viability was measured by an ATP-based assay after 72 hr of drug treatment. (B) Stills from representative time-lapse imaging (three days) of CRC organoids P8T and P26T treated with vehicle (DMSO) or a combination of targeted inhibitors afatinib and selumetinib (both 1 µM) (see also Video 1). In every panel, upper images show color-coded depth of maximum-projected z-stacks of H2B-mNeonGreen fluorescent organoids. Lower panels: corresponding transmitted light images. Time interval: 15 min. Scale bars: 20 µm. Representative time-lapse of two experiments is shown (total six organoids/condition). https://doi.org/10.7554/eLife.18489.003 Figure 1—source data 1 ImageJ/Fiji macro script: 'Organoid movie macro'. Converts XYZT confocal data sets into analyzable 2D-movies, consisting four quadrants: depth coding, maximum projection in 'glow', transmitted light image and a merge between transmitted light and fluorescence. All supplementary movies were generated using this method (Figures 1 and 3 show 2 of 4 quadrants only). https://doi.org/10.7554/eLife.18489.004 Download elife-18489-fig1-data1-v2.ijm To monitor drug response on a cellular level, we stably introduced DNA constructs encoding fluorescently-labeled H2B and performed real-time confocal imaging on the 3D-organoids for 72 hr in the presence and absence of drugs. We performed EGFR-RAS-ERK pathway inhibition with relatively high concentrations of afatinib (1 µM) in combination with selumetinib (1 µM). In P26T (mutant KRAS) we only observed cell cycle arrest with very limited cell death induction. This was in stark contrast with the very rapid induction of cell death in P8T (WT KRAS) (Figure 1B, Video 1). When we repeated these imaging experiments using much lower drug concentrations, we noticed a general shift to resistance for both organoid lines. Under these conditions, also P8T predominantly showed cell cycle arrest rather than cell death, and the cancer cells in P26T organoids even continued to proliferate (Figure 1—figure supplement 1, Video 2). Taken together, our data indicate that 72 hr of combination treatment with afatinib and selumetinib (EGFRi/HER2i and MEKi) effectively kills KRAS WT P8T organoids, while the mutant KRAS P26T organoids are significantly less sensitive. Video 1 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Real-time imaging of cellular drug responses in tumor organoids using high concentrations targeted inhibitors. https://doi.org/10.7554/eLife.18489.006 Video 2 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Real-time imaging of cellular drug responses in tumor organoids using low concentrations targeted inhibitors. https://doi.org/10.7554/eLife.18489.007 In vivo drug response of xenotranplanted patient-derived cancer organoids In order to validate the observed drug response of in vitro cultured organoids in an in vivo model, we xenotransplanted P18T and P26T tumor organoids in immunodeficient mice. In line with a previous report where only engineered tumor progression organoids with increasing number of cancer mutations (APC, KRAS, P53 and/or SMAD4) showed efficient engraftment (Drost et al., 2015), we only obtained reliable engraftment using P26T CRC organoids. We initially started using concentration schedules of afatinib and selumetinib that had previously been reported (Sun et al., 2014), but we observed no significant effect of the drug combination on tumor growth over time (Figure 2A). To exclude that the tumors had acquired resistance during the in vivo drug treatment, we isolated the tumors to re-establish secondary organoids and subjected these to identical drug tests. Dose-response curves on these secondary organoids were identical to the parental organoid line P26T, independent of the type of drug treatment that the tumors underwent in the mice (Figure 2—figure supplement 1). Indeed, in agreement with lower drug concentrations that proved to be ineffective in blocking proliferation in vitro (Figure 1—figure supplement 1, Video 1), we speculate that the in vivo drug concentrations were insufficient to effectively block the EGFR-MEK-ERK pathway. To confirm this hypothesis, we further increased the drug levels to high but tolerable doses. This indeed induced significant growth stabilization (but no regression) of P26T xenotransplanted tumor in mice (Figure 2B), in agreement with loss of proliferative activity as was also detected in vitro (Figure 1B). The fact that in vivo xenografted CRC organoids yields similar drug responses as in vitro organoid cultures and identical to previous reported drug response of KRAS mutant PDX models of CRC (Sun et al., 2014), validates the testing and evaluation of targeted inhibitors in CRC organoids. Figure 2 with 1 supplement see all Download asset Open asset In vivo drug response of xenotransplanted CRC organoids. (A) P26T CRC organoids were subcutaneously transplanted in immunodeficient mice. Once tumors have grown to a volume of 300 mm3, animals were treated for 28 days with vehicle, afatinib (12,5 mg/kg; five days on, two days off), selumetinib (20 mg/kg; five days on, two days off) or both drugs in combination. The mean percentage change in tumor volume relative to initial tumor volume is shown. Error bars represent standard deviation. n.s., not significant. (B) Same experimental setup as in A, but with increased drug concentrations for afatinib (20 mg/kg; five days on, two days off) and selumetinib (25 mg/kg; five days on, two days off); as well as in combined treatment. Error bars represent standard deviation. *p<0,05; **p<0,01; ***p<0001. https://doi.org/10.7554/eLife.18489.008 CRISPR genome-editing in CRC organoids reveals profound effect of KRASG12D on drug response P8T and P26T CRCs are microsatellite-stable (MSS) and belong to the same molecular subtype classification based on RNA expression data (TA, also referred to as canonical CMS2 according to consensus classification) (van de Wetering et al., 2015; Guinney et al., 2015). Genomic characterization of these patient-derived CRC organoids in comparison to their matched normal tissue revealed many additional mutations within the protein coding sequence of the genome (van de Wetering et al., 2015). For P8T and P26T, 230 and 163 of such cancer specific mutations were detected respectively (van de Wetering et al., 2015). To exclude potential contributions of all these additional mutations to the effect that oncogenic KRAS imposes on drug responses, we introduced an oncogenic KRAS mutation in patient-derived CRC organoid P18T via CRISPR/Cas9-mediated homologous recombination (Drost et al., 2015). Like P8T, original P18T is WT for the entire downstream EGFR signaling pathway. P18T-KRASG12D mutant cells were generated as reported previously for normal colon organoids (Drost et al., 2015) and genotyping of clonally expanded organoids confirmed that the clones contained the KRASG12D mutation (Figure 3A), as well as a Cas9-mediated inactivation of the second allele by introducing an 86 bp deletion. Upon addition of oncogenic KRAS, no overall differences in morphology or growth rates were observed during normal culture conditions. Figure 3 with 4 supplements see all Download asset Open asset CRISPR genome editing in CRC organoids reveals effect of KRASG12D on drug response . (A) Schematic representation of the CRISPR/Cas9-induced homologous recombination strategy to introduce the KRASG12D mutation in the KRASWT patient-derived CRC organoid P18T. Green bar: start codon. Red bar: G12D mutation. Parental and mutant sequences are shown on the right. (B) Extensive dual-inhibitor dose-response assay of patient-derived CRC organoids P18T and P18T-KRASG12D treated for 72 hr. 14×14 drug concentrations of afatinib and selumetinib were chosen with logarithmic intervals covering a 5 nM–5 μM range. The results of the full matrix screen are represented as a heat map (left), where red represents 0% ATP levels (no viability) and green represents 100% ATP levels (max viability). The dose-response curves to the right represent the horizontal (afatinib monotherapy), vertical (selumetinib monotherapy) and diagonal (afatinib/selumetinib combination therapy) lines in the heat maps. Dashed lines are P18T; solid lines are P18T-KRASG12D. (C) Stills from representative time-lapse imaging (three days) of CRC organoids P18T and P18T-KRASG12D treated with vehicle (DMSO) or afatinib + selumetinib (both 1 µM) (see also Video 1). In every panel, upper images show color-coded depth of maximum-projected z-stacks of H2B-mNeonGreen fluorescent organoids. Lower panels: corresponding transmitted light images. Time interval: 15 min. Scale bars: 20 µm. Representative time-lapse of 2 (total eight organoids/condition) and four experiments (total 20 organoids/condition) for P18T and P18T-KRASG12D resp. (D) Mitotic and apoptotic events in the organoid drug response movies (C and Video 1) were manually marked and quantified (see Materials and methods and Figure 3—figure supplement 3). In comparison with vehicle (-), drug treatment of p18T with afatinib and selumetinib (a+s) results in both proliferation block and apoptosis induction, while p18T-KRASG12D only shows reduced proliferation but unchanged apoptosis rates. Error bars represent standard deviation. *p<0,05; ***p<0,001; n.s. = not significant (p=0,4) https://doi.org/10.7554/eLife.18489.010 To investigate the exclusive effect of oncogenic KRAS on a combination therapy that targets the EGFR-RAS-ERK pathway, we performed a full matrix screen of 14 drug concentrations over a 5 nM to 5 µM range of both the targeted inhibitors afatinib (EGFR/HER2i) and selumetinib (MEKi) (Figure 3B). Notably, their combined administration is currently used in a clinical trial for patients with RAS mutant CRCs (NCT02450656). While the original P18T demonstrated high sensitivity to EGFR/HER2 inhibition by monotherapy, a single introduced oncogenic point mutation in KRAS provided resistance to EGFR/HER2 inhibition. Moreover, we analyzed combination effects using the Bliss independence model. Positive Bliss scores indicate combinatorial effects that exceed additive effects. The heat map of Bliss scores for P18T and P18T-KRAS shows that a large range of concentrations for both compounds show positive scores, but that presence of oncogenic KRAS renders the loss of viability and positive Bliss range towards higher drug concentrations indicating resistance (Figure 3—figure supplement 1). Next, we again studied the cellular drug response by real-time imaging. Reminiscent of the patient-derived CRC organoid with an endogenous RAS mutation (P26T), we noticed that the introduction of oncogenic KRAS renders a CRC organoid less sensitive to the afatinib/selumetinib combination therapy (Figure 3C, Video 1). More specifically, quantifications of all mitotic and apoptotic events during the filmed drug response revealed both loss of proliferation and apoptosis induction in P18T, while P18T-KRASG12D only showed reduced proliferation but unchanged apoptosis rates (Figure 3D and Figure 3—figure supplement 2). Despite the phenotypic difference in drug response, pERK levels in both tumor organoids were severely reduced (Figure 3—figure supplement 3). Since suboptimal suppression of ERK activity might permit tumor growth in BRAF mutant cancers (Bollag et al., 2010; Corcoran et al., 2015), we determined the cellular effects of drug response when lowering drug concentrations. Since significant differential effects were observed between P18T and P18T-KRASG12D during the matrix screen around 33 nM afatinib + 200 nM selumetinib (Figure 3B), we repeated real-time imaging of drug response using these lower drug concentrations. As with P8T and P26T, we noticed a general shift from sensitivity towards resistance when drug concentrations were reduced. More specifically, the RAS WT cancer organoids showed cell cycle arrest rather than cell death, while the RAS mutant organoids appeared unaffected and sustained proliferation (Figure 3—figure supplement 4, Video 2). Differential drug sensitivity in CRC organoids with and without mutant RAS upon combination therapies that include EGFR inhibition Considering the isogenic CRC organoids P18T and P18T-KRASG12D as our gold standard to reveal the specific effects of KRASG12D on drug responses, we expanded our focus at targeting the linear EGFR-RAS-ERK pathway with the ultimate aim to find a targeted therapy that is specifically effective against RAS mutant CRCs. Multiple targeted inhibitors against identical targets were used to exclude artifacts and to increase the mechanistic significance behind the rationale of potential therapies (Figure 4A; and Figure 4—source data 1 and Supplementary file 1 for all dose-response curves) of which few combination therapies are in clinical trial (Figure 4B). Figure 4 with 2 supplements see all Download asset Open asset Differential drug sensitivities upon combination therapies including EGFR inhibition. (A) Heat map of dose-response measurements (cell viability) in CRC organoids P18T (top panel) and P18T-KRASG12D (bottom panel). Organoids were treated (72 hr) with vehicle (DMSO) or inhibitors targeting the EGFR-RAS-ERK pathway (5 nM – 20 μM range, in 22 logarithmic intervals). Red represents 0% ATP levels (max cell death) and green represents 100% ATP levels (max viability). Drug names and their nominal targets are indicated in the left panel. Combination therapies that are currently in clinical trial for patients with RAS mutant CRCs are indicated in red font. See Figure 4—source data 1 and Supplementary file 1 for all dose-response curves. (B) Dose-response curves of CRC organoids P18T (dashed lines) and P18T-KRASG12D (solid lines) treated with combination therapies that are currently in clinical trial for patients with RAS mutant CRCs. https://doi.org/10.7554/eLife.18489.015 Figure 4—source data 1 Dose-response curves for patient-derived tumor organoids P18T and P18T KRASG12D as indicated. A number of biological for dose-response are indicated between second combination Download Figure 4—source data 2 Dose-response curves for patient-derived tumor organoids P18T and P18T KRASG12D as indicated. A number of biological for dose-response are indicated between second combination Download we noticed a much lower sensitivity of P18T-KRASG12D for pan-HER inhibitors afatinib, and in contrast to the parental P18T lower Figure 3—figure supplement 1). within P18T additive sensitivity could be observed when was with MEK or ERK inhibition (Figure supplement 1). In strategies strongly in P18T-KRASG12D which specific inhibitor combination was used (Figure supplement 1). all tested combinations that EGFR inhibition revealed effect on cellular viability in P18T than in P18T with mutant KRAS Figure supplement 1). In and combination therapies against MEK and/or ERK that showed on similar in P18T-KRASG12D as in P18T (Figure supplement 1). In we tested strategies and EGFR-RAS-ERK the between these (Figure supplement 2A). Like MEK or ERK inhibition, we observed that inhibition of or in combination with therapy not efficacy in a KRAS mutant (Figure supplement In line with clinical studies on MEK inhibitors with or inhibitors in KRAS mutant CRCs not results et al., 2012). to targeted inhibition of the EGFR-RAS-ERK pathway are in normal and tumorigenic organoids Next, we to further establish whether the effects of oncogenic KRAS on drug response is on a tumorigenic or could independent of cellular We therefore used normal colon organoids and a of that line in which the oncogenic KRASG12D mutation was introduced via similar CRISPR/Cas9-mediated genome-editing strategy as in P18T (Drost et al., 2015). In with mouse studies et al., 2014), we observed no induction of upon introduction of oncogenic KRAS (Figure supplement 1). drug response of normal organoids to targeted inhibitors against the EGFR-RAS-ERK pathway (Figure 5 and Figure supplement revealed a similar as in CRC organoid P18T (Figure 4 and Figure supplement 1). the effect that oncogenic KRAS imposes on drug response independent of cellular and the presence of additional cancer Figure 5 with 2 supplements see all Download asset Open asset drug response in normal and tumorigenic (A) Heat map of dose-response measurements of cell viability in normal colon organoids (top panel) and in normal colon organoids with an oncogenic KRAS mutation (bottom panel) after 72 hr drug treatment with inhibitors targeting the EGFR-RAS-ERK pathway. Same concentration range and as in Figure Combination therapies that are currently in clinical trial for patients with RAS mutant CRCs are indicated in See Figure data 1 and Supplementary file 1 for all dose-response curves. (B) Dose-response curves of normal organoids (dashed lines) and normal organoids + KRAS (solid lines) treated with combination therapies that are currently in clinical trial for patients with RAS mutant CRCs. Figure data 1 Dose-response curves for normal and normal KRASG12D organoids as indicated. of biological for dose-response are indicated between second combination Download a panel of human CRC organoids the differential effect of EGFR inhibition Next, we to our towards a of CRC organoids that is representative for the We additional patient-derived CRC organoids for combinatorial therapies against the EGFR-RAS-ERK signaling pathway. Since all the organoid lines are in of genome we could CRC organoids with and without a mutant RAS pathway. on EGFR/HER2 dual inhibition by afatinib, we could the organoid panel in two types of the sensitive the resistant (Figure green red lines Indeed, drug sensitivity towards all tested EGFR inhibitors correlated with the of KRAS. However, there were two (Figure and Figure supplement 1). The was WT for KRAS, to an oncogenic mutation in the resistant The second was organoid line the other CRC organoids in our panel, is as including the (van de Wetering et al., 2015). tumor contains a BRAF resistance towards the targeted drugs et al., these two that drug screening on human organoids is to the of entire oncogenic the of the frequent Figure with 1 supplement see all Download asset Open asset multiple human CRC organoids confirm RAS for EGFR inhibition. (A) Dose-response curves of patient-derived CRC organoids and one engineered CRC organoid