This protocol describes a standardized evaluation of drug sensitivities to targeted signaling inhibitors in NSCLC patient-derived organoid models.
Novel 3D cancer organoid cultures derived from clinical patient specimens represent an important model system to evaluate intratumor heterogeneity and treatment response to targeted inhibitors in cancer. Pioneering work in gastrointestinal and pancreatic cancers has highlighted the promise of patient-derived organoids (PDOs) as a patient-proximate culture system, with an increasing number of models emerging. Similarly, work in other cancer types has focused on establishing organoid models and optimizing culture protocols. Notably, 3D cancer organoid models maintain the genetic complexity of original tumor specimens and thus translate tumor-derived sequencing data into treatment with genetically informed targeted therapies in an experimental setting. Further, PDOs might foster the evaluation of rational combination treatments to overcome resistance-associated adaptation of tumors in the future. The latter focuses on intense research efforts in non-small-cell lung cancer (NSCLC), as resistance development ultimately limits the treatment success of targeted inhibitors. An early assessment of therapeutically targetable mechanisms using NSCLC PDOs could help inform rational combination treatments. This manuscript describes a standardized protocol for the cell culture plate-based assessment of drug sensitivities to targeted inhibitors in NSCLC-derived 3D PDOs, with potential adaptability to combinational treatments and other treatment modalities.
Personalized therapies against oncogenic drivers have revolutionized cancer treatment, improving patient survival and reducing treatment-mediated side effects1. Recent advances in molecular diagnostics and sequencing technologies have highlighted the complexity of human tumors, with spatial and temporal heterogeneity impacting treatment response2. Recapitulating these subclonal differences in cell culture models has long been limited to investigating selected alterations of interest in otherwise uniform cell lines. Newly developed 3D PDO models generated from tumor biopsies or surgical tumor resections allow for improved representation of cellular complexity and signaling crosstalk within patient-derived tumor tissue3. As such, tumor organoids derived from gastrointestinal and pancreatic cancer have successfully been generated and recapitulate the genetic diversity and determinants of treatment response4,5,6. In non-small cell lung cancer (NSCLC), organoid development and establishment challenges are acknowledged, and optimization of culture techniques and selective media factors is needed to enable broader and more systematic use of NSCLC PDOs in the future7,8.
Developing combinatorial therapies targeting residual tumor cells that withstand initial drug treatment is essential to inhibit resistance development and ultimately to improve patient survival9. Given the architectural complexity of organoid cultures, classical drug response parameters need to be optimized to allow for accurate and reproducible testing of drug sensitivities. Imaging-based readouts10,11 and classical cell viability assays measuring cellular ATP abundance6,12, amongst other techniques, are available to profile drug responses in PDO cultures. Here, we develop and describe a standardized protocol to evaluate drug sensitivities to targeted therapy against known clinical drivers in NSCLC PDO models.
For human subjects research, informed consent was obtained and tissue collection was carried out under the UCSF Internal Review Board approved protocols (IRB, protocol no.: #13-12492, or CC#17-23309). The establishment of organoid cultures from de-identified clinical specimens was performed in collaboration with research partners according to previously published methods13,14,15,16. Organoid cultures were retrieved for maintenance and drug escalation experiments at passage three or later. All the following protocols were performed under aseptic conditions in a mammalian tissue culture laboratory environment.
Figure 1: Protocol schematic of workflow and critical steps in the technique. (A) Experimental workflow including seeding of organoids in 96-well format, treatment with drug escalation at 7 days after seeding, and luminescence-based cell survival readout 5 days after treatment using an ELISA plate reader. (B) An example image of the EGFRdel19-positive TH107 and EGFRL858R-positive TH330 NSCLC organoid cultures. Original cultures, cells at the time of seeding (day 0), and organoids at treatment start 7 days after seeding (day 7) are shown. Scale bar = 100 µm. Changes in organoid diameter over the initial 7-day culture period are quantified and indicate >2-fold increase in organoid size. Relative fold changes in sizes at day 7 compared to average size at day 0 are presented below the representative images. For TH107, a fold change of 2.38 over 7 days is observed, indicating a doubling time of 5.88 days (141.12 h). For TH330, a fold change of 2.41 over 7 days is observed, indicating a doubling time of 5.81 days (139.42 h). Quantification of changes in organoid size and statistical evaluation are presented (right). Statistical significance is calculated by unpaired t-test, p < 0.0001. (C) Treatment layout for drug escalation in organoid 96-well plate format. The number of technical replicates and exemplary doses are indicated, including a negative control. The schematics are created with BioRender, a web-based illustration tool. Please click here to view a larger version of this figure.
1. Experimental preparations
2. Generating single-cell suspension and seeding of cells
3. Drug treatment
4. Readout by luminescence-based survival assay
Considerable challenges in establishing NSCLC organoids have been noted7. Thus, it is exciting to see recent work establishing lung cancer organoids and using them for drug treatment assays20,21,22. EGFR-mutations account for 11.3% of NSCLC cases23. Targeted treatment with EGFR inhibitors represents the first-line treatment option in EGFR-mutant NSCLC and has improved the overall survival and treatment safety in patients24. This work determined the sensitivity to the FDA-approved EGFR tyrosine kinase inhibitor osimertinib24,25 in EGFR-mutant NSCLC organoids. EGFR-mutant NSCLC organoids were generated from surgical resection or tumor biopsy specimens of NSCLC patients and confirmed to harbor the indicated oncogenic mutation by DNA sequencing. As outlined above, EGFR-mutant NSCLC organoid models were treated with escalating doses of osimertinib and PDO viability assessed by luminescence-based cell survival readout five days after the treatment initiation. While EGFR mutant (EGFRdel19)-positive TH107 organoids showed sensitivity to osimertinib treatment with a half-maximal inhibitory concentration (IC50) of 56 nM (Figure 2A), EGFR-mutant (EGFRL858R)-positive TH116 organoids were resistant to osimertinib treatment with an IC50 of greater than 1 µM (Figure 2B). The sensitivity of EGFRdel19-positive TH107 NSCLC was accompanied by significant transcriptional changes, including a reduction in the expression of cell cycle-associated gene signatures and an increase in the expression of apoptosis-associated gene signatures (Supplementary Figure 2A,B). As a reference, response data for the sensitive EGFRdel19-positive NSCLC cell line PC9 is presented (Figure 3A,B). The latter includes survival analysis to escalating doses of osimertinib by a 2D luminescence-based survival assay (Figure 3A) and the study of signaling suppression on the level of EGFR-MAPK signaling by Western blot (Figure 3B). Overall, this data highlights the accuracy of the present protocol for determining drug response and distinguishing between sensitive and resistant NSCLC PDO models. Further analyses of EGFRL858R-positive TH116 organoid and available clinical specimens are needed to determine possible resistance-associated alterations.
Figure 2: Treatment response curve of EGFR-mutant NSCLC organoid models to osimertinib escalation. (A) Osimertinib response in the sensitive EGFRdel19-positive TH107 NSCLC organoid model. (B) Osimertinib response in the resistant EGFRL858R-positive TH116 NSCLC organoid model. Data points are presented as normalized values showing the mean +/- standard deviation, with a non-linear regression curve fitted through the data. TH107, n = 6 technical replicates per data point. TH116, n = 4 technical replicates per data point. Please click here to view a larger version of this figure.
Figure 3: Comparative data for the treatment response to osimertinib in a sensitive EGFR-mutant NSCLC cell line and organoid models cultured in different media. (A) Osimertinib response in the sensitive EGFRdel19-positive NSCLC cell line PC9, determined by standard 2D-CTG assay. (B) Signaling suppression in the PC9 cells upon two-day treatment with osimertinib (2 µM). (C) Osimertinib response in the sensitive EGFRL858R-positive TH330 NSCLC organoid model culture in LGM and GM media. (D) Osimertinib response in the resistant AZ021 NSCLC organoid model in LGM and GM media. Confirmation of the oncogenic EGFRL858R mutation in AZ021 failed and may be causative for the lack of osimertinib response. For A and C-D, data points are presented as normalized values showing the mean +/- standard deviation, with a non-linear regression curve fitted through the data. PC9, n = 3 technical replicates per data point. TH330, n = 5 technical replicates per data point. AZ021, n = 6 technical replicates per data point. A Wilcoxon rank test was performed on normalized data to determine the statistical significance. For TH330 (C), LGM vs. GM, ** p = 0.0078. For AZ021 (D), LGM vs. GM, ns p = 0.0742. Please click here to view a larger version of this figure.
Supplementary Figure 1: Representative cell analyzer results for counting and viability assessment of EGFR-mutant organoid models. TH107 and TH107BC refer to different organoid models and A and B to biological replicates. For each model and biological replicate, three technical replicates are counted; all show ≥95% viability. A representative image during cell counting is presented on the right, showing robust viability and single-cell dissociation. Please click here to download this File.
Supplementary Figure 2: Gene set enrichment analysis (GSEA) using bulk RNA sequencing data obtained for EGFR-mutant TH107 NSCLC organoids, comparing untreated control (DMSO) and cells treated with Osimertinib for 3 days (OSI_D3). (A-B) Upon targeted treatment, sensitive cells will undergo cell cycle G1 arrest and cease active proliferation. As an expression of G2M cell cycle genes is associated with active proliferation, a nominal enrichment of expression in the untreated control (DMSO) is expected. For apoptosis-related genes, a nominal enrichment in treated cells (OSI_D3) is expected. Both have been confirmed in the EGFR-mutant TH107 NSCLC organoid model treated with Osimertinib: (A) GSEA for Hallmark G2M expression signature (left) shows enrichment in DMSO- treated cells. Nominal enrichment score (NES): +1.708, FDR < 0.0001. (B) GSEA for Hallmark Apoptosis expression signature (right) shows enrichment in Osimertinib-treated cells. NES: -1.075, FDR: ns, 0.3275. Please click here to download this File.
Supplementary Figure 3: Combinatorial drug treatment in EGFR-mutant TH330 organoid model treated with Osimertinib escalation in the presence of a second additive inhibitor at a fixed concentration. Resistance-associated alterations in EGFR-mutant NSCLC, i.e., SRC and AXL activation26,27,28, were pharmacological targeted by combinatorial treatment with SRC inhibitor Saracatinib (100 nM) or AXL inhibitor R428 (500 nM), n = 6 technical replicates per data point. Both combinatorial treatments resulted in increased treatment response, with significance for the combination of Osimertinib with SRC inhibitor Saracatinib. Statistical significance was evaluated by Wilcoxon rank test: Osimertinib versus Osimertinib + Saracatinib, *p = 0.0195; Osimertinib vs. Osimertinib + R428, ns, p = 0.2500. Please click here to download this File.
Supplementary Table 1: Change of organoid size from seeding (d0) to 7 days post-treatment (d7). (A) Growth development in EGFRdel19-positive NSCLC organoid TH107. (B) Growth development in EGFRL858R-positive NSCLC organoid TH330. Please click here to download this Table.
This manuscript develops and describes a standardized protocol for assessing drug sensitivity in NSCLC-derived 3D PDO models. In addition to drug sensitivity studies, further characterization of available organoid models is needed to determine the underlying causes for differences in drug sensitivity. This may include genetic profiling of organoids and patient specimens and other analysis available for organoids, such as immunohistochemistry staining for differentiation markers and general cellular signaling biomarkers and physiology13,29.
Critical steps in the protocol
The protocol outlined herein provides a standardized workflow that allows accurate and reproducible drug sensitivity analyses when followed carefully. Particular care should be taken in the following steps: TrypLE and DNAse I digestion during generation of single-cell suspensions, seeding of single-cell suspensions in BME2, monitoring organoid growth until treatment, media changes, and disruption and lysis of BME2 embedded organoids during luminescence-based cell survival readout. (1) While additional DNAse I digestion after TrypLE-based dissociation of organoids is not essential for expanding organoid models during regular culture maintenance, DNAse I digestion should not be omitted when seeding for drug escalation experiments as it ensures better separation of organoid clusters into single-cell suspensions and accurate cell counting. (2) Seeding of single-cell suspension in BME2 represents a critical step given the solidification of BME2 at room temperature. Thus, a maximum of 1-2 rows needs to be seeded at once, and samples should be placed on ice before additional rows are seeded. Of note, cells need to be pipetted up and down when seeding is continued to allow for a homogeneous cell suspension. (3) Organoid growth needs to be monitored carefully during the 7-day expansion from seeding to treatment. An example of the expected development is given in Figure 1B and Supplementary Table 1. Of note, assessing changes in organoid size by brightfield microscopy and image analysis as presented in Figure 1B may allow for a precise evaluation of differences in organoid growth and doubling times. Doubling times can have an impact on drug responses, as recently discussed in the literature30. If the organoid growth rate exceeds the presented example significantly, a shorter expansion time until the start of treatment and shorter treatment duration can be considered. (4) In addition, special care should be taken when changing media to avoid aspirating organoids. The seeding position of BME2 embedded organoids at the 6 o'clock position allows for a safe aspiration of media when plates are turned clockwise by 180° and media is aspirated at the opposite position of the organoids. (5) Finally, thorough lysis of BME2 embedded organoids during the survival readout is essential to record accurate results. According to the manufacturer's instructions, samples should be pipetted up and down repeatedly, ideally using unfiltered tips, to ensure proper lysis. Incubation times should be followed as described. Further, transferring 75% of the lysate (instead of the total volume) to a white, opaque-bottom 96-well plate for the final readout using an ELISA plate reader allows for an appropriate assessment, as this assures the same volume in each well and the absence of air bubbles that can be introduced by vigorous pipetting.
Of note, profiling of drug responses in BME2-embedded organoid cultures may show a higher standard deviation than observed in regular cell line cultures (Figure 2, Figure 3A). The higher standard deviation is based on several factors, including an increased likelihood of minor variations in seeding when working with BME2 and differences in individual organoid growth rates across wells over the initial 7-day growth period. Thus, equal or more than four technical replicates per drug concentration should be seeded.
Most importantly, the presence of malignant cells carrying the oncogenic driver mutation and limited contamination by normal airway epithelial cells must be carefully evaluated. Challenges in NSCLC establishment can favor the outgrowth of normal airway epithelial cells7. Copy number profiling or PCR- and sequencing-based approaches to confirm the presence of the oncogenic driver mutations are the methods of choice to ensure the quality of NSCLC organoid cultures.
Modifications and troubleshooting of the method
Media and respective growth factors added to basic media solutions can significantly impact drug response to targeted inhibitors. They activate bypass receptors and signaling pathways that influence and limit drug response (e.g., FGF, HGF, EGF)26. While a growth-factor rich and tailored media may be optimal for expanding organoid culture, drug escalation and sensitivity assessments should be performed in a reduced growth-factor media, as outlined above. This is based on internal experience comparing different media formulations and drug response data (Figure 3C). While media solutions can affect the degree of sensitivity to certain drug treatment and can shift IC50 values, robust phenotypes of sensitivity or resistance are apparent irrespective of media formulation (Figure 3C,D). In addition, general consistency in the media formulation and profiling drug responses across organoid cultures is recommended, and an equal or more than four technical replicates per concentration needs to be seeded. This is particularly important to benchmark ranges in sensitivity vs. resistance for the Inhibitor of interest.
Limitations of the method
The protocol presented here describes the sensitivity of NSCLC 3D cancer organoid models to targeted inhibitors when patient-derived cancer cells are cultured. Additional experiments, including pharmacodynamic analysis regarding pathway inhibition and sequencing analysis for the presence of the driver oncogene and secondary mutations, are needed for a detailed characterization of drug resistance and sensitivity. Further, bystander factors such as microenvironmental stimuli derived from interactions or secreted factors by non-cancer bystander cells in the tumor microenvironment are not accounted for, and novel protocols are needed when co-culture organoid models with immune or stromal cells are attempted. Recent work has highlighted the use of organoid models to recapitulate tumor microenvironment interactions and profile responses to immune checkpoint inhibitors, such as anti-PD-L1 treatment13,31.
The significance of the method with respect to existing / alternative methods
3D cancer organoid models recapitulate the genetic diversity and determinants of treatment response present in the original tumor4,5,6. Notably, spatial and temporal heterogeneity can promote tumor evolution, and parallel emergence and the sequential development of tumor subclones can occur32,33. Intratumor heterogeneity is significant for the selection of more resilient tumor cells under therapeutic pressure9,34,35. The protocol provided here allows for a rapid assessment of sensitivities to treatment with targeted inhibitors in patient-proximate samples. Thus, organoid models have advantages over more conventional homogeneous cell line models lacking genetic diversity or long-term studies using cell lines or patient-derived xenografts. Further, the present protocol allows scaling up to multiple arms of treatment and combinational treatment approaches with few limitations regarding cost and analytic capacity. As such, adding a second drug of interest at a fixed dose while escalating the primarily targeted Inhibitor and comparing it to the escalation of the primarily targeted Inhibitor alone allows for efficiently evaluating the potential combinatorial effects and with minimal additional biomass required (Supplementary Figure 3). Compared to imaging-based assessments used to monitor organoid development and drug response, the luminescence-based cell survival assay described here has similar sensitivity with minimal equipment and training required.
Importance and potential applications of the method in specific research areas
Developing a standardized pipeline that allows for establishing cancer organoid models from patient specimens and the subsequent drug sensitivities profiling holds significant clinical applicability potential. Ex vivo pharmacological profiling has gained recognition in detecting vulnerabilities and resistant-associated features in tumors, correlating to treatment response in patients36,37. Significantly, ex vivo profiling of drug sensitivities may aid in treatment selection in the clinic and the design of rational combinational treatments addressing resistance mechanisms. Overall, this approach could help to enable improved personalized strategies for molecular therapy or combinatorial treatment regimens. The latter may help target drug tolerance and resistance mechanisms early and deepen clinical response to improve patient outcomes in the future.
The authors have nothing to disclose.
We thank the laboratories of Jeroen P Roose (UCSF) and Calvin J Kuo (Stanford) for their input regarding organoid culture and protocol development. We further thank Oghenekevwe M. Gbenedio (Roose lab, UCSF) for protocols and sample establishment input. This research project was conducted with support from the NIH [U54CA224081]. F. Haderk was supported by the Mildred Scheel postdoctoral fellowship from the German Cancer Aid.
1.5 mL tubes | |||
15 mL centrifuge tubes | |||
500 mL Vacuum Filter/Storage Bottle System, 0.2 µm Pore 33.2 cm2 Nylon Membrane | Corning | 430773 | for both media |
96-Well, Cell Culture-Treated, Flat Clear Bottom Black Microplate | Corning | 3904 | |
96-Well, Cell Culture-Treated, Solid White Flat-Bottom Microplate | Corning | 3917 | |
A-8301 | Tocris Bioscience | 293910 | for both media |
Advanced DMEM/F-12 | Gibco | 12634010 | for LGM |
B27 | Life Technologies | 12587010 | for both media |
BioRender 2021 | https://biorender.com/ | online scientific illustration software | |
BME2 (Cultrex RGF Basement Membrane Extract, Type 2) | R&D Systems | 353301002 | |
Cell culture incubator (37 °C, 5% CO2) | |||
CellTiter-Glo 3D Cell Viability Assay | Promega | G9682 | 3D-CTG readout reagent |
Centrifuge holding 15 mL centrifuge tubes | |||
Deoxyribonuclease I (DNAse I) | ThermoFisher Scientific | 18047019 | |
Dulbecco's Phosphate-Buffered Salt Solution | Corning | MT21031CV | |
DMEM/F-12, GlutaMAX supplement | Gibco | 10565018 | for GM |
GlutaMax | Gibco | 35050061 | for LGM |
GraphPad Prism software (version 9.2.0) | GraphPad | statistical analysis software | |
HEPES | Gibco | 15630080 | for both media |
hFGF-10 | PeproTech | 100-26-100ug | for GM |
hFGF-7 | PeproTech | 100-19-50ug | for GM |
hNoggin | PeproTech | 120-10C-100ug | for both media |
hRspondin | PeproTech | 120-38-100ug | for GM |
Low retention pipette tips, 20 µL (P20) | ThermoFisher Scientific | 2149P-05-HR | |
Low retention pipette tips, 200 µL (P200) | ThermoFisher Scientific | 2069-05-HR | |
Regular length pipette tips, 1000 µL (P1000) | ThermoFisher Scientific | 2179-HR | |
Multichannel pipette | |||
N-Acetylcysteine | Fisher Scientific | 50-424-777 | for both media |
Nicotinamide | Sigma Aldrich | N0636-100G | for both media |
Osimertinib | Selleck Checm | S7297 | |
Penicillin-Streptomycin-Glutamine | Gibco | 10378016 | for LGM |
Penicillin/Streptomycin | Cytiva HyClone | SV30010 | for GM |
Pipettes (different sizes) | |||
Plate reader | Molecular Devices | SpectraMax M5 | equipment, alternative readers may be used |
Primocin | Invivogen | ant-pm-1 | for GM |
SB202190 | Selleck Chem | S1077 | for GM |
TrypLE Express Enzyme | Gibco | 12604021 | |
Vacuum pump and tubing | |||
Vi-CELL XR Cell Analyzer | Beckman Coulter | Vi-CELL XR | cell analyzer / counter |