High-Content Clonogenic Survival Screen to Identify Chemoradiation Sensitizers

Abstract

The combination of cytotoxic chemotherapy with radiotherapy (CRT) has resulted in significant improvements in clinical outcomes for patients with many locally advanced unresectable cancers. Only a small proportion of patients achieve pathological complete responses to CRT; combination of CRT with targeted agents offers the promise of further improving treatment responses. However, numerous clinical trials have failed to show an improvement in clinical outcomes with the addition of targeted agents. In part, the gap in translation to large, expensive and ultimately unsuccessful clinical trials is a reflection of the shortcomings of inconsistently designed, executed and reported preclinical data that these studies are based on. In an effort to standardize the selection of agents for future clinical testing, we have designed, optimized and validated a high throughput clonogenic assay platform for step-wise progression of preclinical studies from in vitro to in vivo in non-small cell lung cancer (NSCLC) and pancreatic ductal adenocarcinoma (PDAC). This highly stable in vitro method was standardized for identification of the most promising drugs that could best be combined with CRT from amongst as screen of multiple agents tested in an unbiased manner using 96-well plates. The methodology lends itself to seamless testing of multiple agents in a similar fashion allowing cross-comparisons, evaluation of CRT and/or radiation therapy alone, and testing multiple concentrations of test agents sequenced at different times before and after radiation. The method identified Trametinib (MEKi) as a strong CRT sensitizer in KRAS-mutant NSCLC and PDAC cell lines. To increase the accessibility of our screening method and accelerate the pace at which novel combinations with CRT are identified and incorporated into standard practices for treatments, we report details on screening method optimization, data generation, and downstream data analysis.

Introduction

Chemoradiation (CRT) has become an indispensable component of treatment for multiple solid tumors in neoadjuvant, primary, or adjuvant setting in the past few decades [1]. It has become the standard of care for most locally advanced unresectable cancers and has significantly improved the clinical outcomes for these patients [2, 3]. A major rationale for combining chemotherapy and radiotherapy is the radio-sensitization effect of chemotherapeutic agents [4]. Integration of targeted agents into the CRT regimen offers the promise of further improving treatment efficacy [5, 6]. However, biological factors such as the intrinsic and extrinsic radioresistance mechanisms in tumor cells and inter- and intra-tumoral heterogeneity are the major barriers towards maximizing the effectiveness of radiotherapy [7]. Therefore, there is a great need to identify novel CRT sensitizers that improve the treatment efficacy while not increasing toxicity for the patients.

Despite promising preclinical results demonstrating additive or supra-additive effects with the addition of targeted agents to CRT, numerous clinical trials have failed to show an improvement in clinical outcomes with the addition of targeted agents [8]. From the standpoint of in vitro testing, the key contributors to this disconnect between preclinical promise and clinical failure are the frequent restriction of preclinical studies to combination of targeted agent and radiation (RT) alone and not CRT, the lack of standardization of radiation dosimetry and protocols (addressed elsewhere in this special issue), and the insufficient evaluation of optimal sequencing strategies. To overcome the lack of standard procedures in preclinical development of CRT sensitizers, we developed a reproducible high-throughput screening methodology that evaluates cells treated with novel targeted therapies administered before or after standard-of-care chemotherapy and radiation therapy. We employed our unbiased, systematic preclinical approach in non-small cell lung cancer (NSCLC) and pancreatic ductal adenocarcinoma (PDAC) to screen the portfolio of Cancer Therapy Evaluation Program (CTEP) agents for CRT sensitizers.

The current gold standard for assessing the CRT sensitivity of a cell line and the effectiveness of agents thought to modify CRT effects is still the clonogenic survival assay (CSA), tested in 6-well plates [9]. However, the low-throughput nature of CSA has limited its ability to be used as the standard read-out in large-scale drug screening studies. Hence, a few alternative techniques such as MTS [10, 11], γH2AX foci [12], double strand break fluorescent report system [13], and syto60 nucleic acid staining [14], that measure the short-term survival or proliferation of cell lines or DNA repair capability of cells have been utilized as metabolic or colorimetric surrogates for radiosensitivity in high-throughput experiments using multi-well formats (24-well plate, 96-well plate, and 384-well plate). While these methods facilitate increasing the number of drugs to be tested in one experiment, they do not reflect the long-term survival of cell lines after treatment or their ability to form colonies, an indication of preserved reproductive capacity after genotoxic stress. Adopting and adapting the High-throughput clonogenic screening assay (CSA) method we published previously [Lin 2014], we now describe the utility of such an unbiased high-throughput screening method to identify CRT sensitizers among hundreds of drugs within one experiment in 96-well plate format with standard CSA read-outs by utilizing non-small cell lung cancer (NSCLC) and pancreatic ductal adenocarcinoma (PDAC) cell lines. To facilitate the repeatability and standardization of our screening method, we report detailed protocols regarding on screening system optimization, cell line selection, sequence of treatment, experimental designs, and data processing and analysis.

Materials and Methods

Cell Lines

All of the NSCLC cell lines were either obtained from the American Type Culture Collection (ATCC) or provided by Dr. John Minna (University of Texas Southwestern Medical Center). All cell lines were maintained in RPMI 1640 + 2mM L-glutamine + 10% FBS at 37 °C with 5% carbon dioxide in the tissue culture incubator. All the cell lines are tested for mycoplasma before use for experiments.

Reagents and Irradiator

Trametinib and Vorinostat (SAHA) were purchased from Selleck Chemicals. All the CTEP drugs were obtained from CTEP portfolio. Paclitaxel, carboplatin and capecitabine were purchased from MD Anderson Pharmacy. Cells were irradiated on an orthovoltage XRAD320 irradiator (Precision X-Ray). The irradiator was calibrated yearly using dosimetry that is directly traceable to the national institute of standards and technology (NIST) standard.

NSCLC Cell Lines Colony Formation Speed Test and Colony Score Assignment

For each cell line, single cell suspensions were seeded at 100–400 cells/well into 96-well plates with total volume of 100ul. The cell plates are maintained in tissue culture incubator at 37 °C with 5% carbon dioxide undisturbed for 5 – 10 days until cells form a reasonable number of colonies comprised of more than 50 cells as determined by low-magnification microscopy. Afterwards, cells were fixed and stained with methanol-crystal violet (Sigma HT90132) working solution (methanol:crystal violet=9:1) for 15 minutes at room temperature. The wells were then washed with warm tap water 3 to 4 times. The stained 96-well plates were left in hood until they were completely dried (usually overnight). Based on criteria on levels of isolation, circularity, and size of the colonies, we manually assigned a colony formation score of 1–5 where 1 stands for worst score and 5 stands for best score for the formed colonies from each cell line.

Screening test with Trametinib and Vorinostat

Single cell suspensions from H460, H1437 and H358 were seeded at three different densities (100 cells /well, 200 cells/well, and 400 cells/well) with triplicates into 96-well plates with total volume of 90μl. 18 hours after cell seeding, trametinib and Vorinostat at 10-fold the final concentration (Trametinib: 125 nM; Vorinostat: 10 μM) were added at 10 μl per well into the 96-well plates so that the cells are treated with 12.5 nM Trametinib or 1 μM Vorinostat. Within 4 hours after the drug treatment, cells are treated with 2Gy, 3 Gy, and 4 Gy using the XRAD320 irradiator. After the treatment, 96-well plates were put back to tissue culture incubator at 37 °C with 5% carbon dioxide undisturbed for additional 6 days. The cells are then stained with crystal violet and colonies were automatically counted by the IN-Cell Analyzer 6000 (GE Healthcare).

Automated High Content Imaging Analysis

The detailed protocols of the automated high throughput microscopy imaging analysis with IN-Cell Analyzer 6000 were described previously [15]. Briefly, the imaging analysis could automatically identify individual cell nuclei (cell mask) and individual colonies (colony mask). The analysis algorithm then enumerated the total number of cells within each identified colonies consisting of 50 or more cells. The overlapping colonies were only counted as one single colony.

CRT Sensitizers Screening Experiment

Single cell suspensions from H460, H1437, and H727 were seeded at 100 cells/well, 150 cells/well, and 200 cells/well, respectively into 96-well plates in triplicates. For cell plates only treated with radiation, the total volume was 100 μl/well at seeding and for cell plates were treated with CTEP drug and Chemo drug combination, the total volume was 80 μl/well. CTEP drug plates with 6 different concentrations at 10-fold of final concentration (1nM, 10nM, 100nM, 1μM, 10μM, 100μM, 1mM) and chemotherapy drug plates at 10-fold of final concentration (Paclitaxel: 0.0175 μM; Carboplatin: 0.0175 μM) were prepared in advance. After cells were seeded for 18 hours, cells were treated as planned (either radiation only or radiation + CTEP drug + chemotherapy) with 6 different treatment sequences. The 6 treatment sequences were planned as following: 1) Cells are irradiated 1 hour after Chemotherapy and targeted drugs are added (DR_1hr); 2) Cells are irradiated 2 hours after Chemotherapy and targeted drugs are added (DR_2hr); 3) Cells are irradiated 2 hours after Chemotherapy and targeted drugs are added (DR_4hr); 4) Chemotherapy and targeted drugs are added 1 hour after cells are irradiated (RD_1hr); 5) Chemotherapy and targeted drugs are added 2 hours after cells are irradiated (RD_2hr); 6) Chemotherapy and targeted drugs are added 4 hours after cells are irradiated (RD_4hr). The drugs were dispensed from the drug plates into 96-well plates by a 96 multi-channel robotic handler (ViaFLO96, Integra). After the treatment, the 96-well plates were immediately placed back in the tissue culture incubator at 37 °C with 5% carbon dioxide for an additional 5–7 days. Before proceeding to stain the whole plates, the extra negative control wells were firstly stained to be manually checked whether the number of cell colonies formed was ideal. After the check was done with no problem, cells were stained with crystal violet, washed, dried completely, and the colonies were enumerated automatically by IN Cell Analyzer 6000 (GE Healthcare).

Data Analysis

Only colonies with 50 cells or more were counted as colonies. Standard deviation of the colony number among triplicates was calculated for each tested condition. Only conditions that without clear outliers (three standard deviations away from the mean) were included for downstream analysis. For each tested condition, the SRF_2Gy+Chemo value was calculated according to the following formula:

$$\mathbf{SR}{\mathbf{F}}_{\mathbf{2}\mathbf{Gy}\mathbf{+}\mathbf{Chemo}}\mathbf{=} \mathbf{\left(}\mathbf{NC}{\mathbf{N}}_{\mathbf{2}\mathbf{Gy}}\mathbf{x} \mathbf{NC}{\mathbf{N}}_{\mathbf{chemo}\mathbf{+}\mathbf{drug} \mathbf{at} \mathbf{concentration} \mathbf{x}}\mathbf{\right)}\mathbf{/}\mathbf{NC}{\mathbf{N}}_{\mathbf{chemo}\mathbf{+}\mathbf{2}\mathbf{Gy}\mathbf{+}\mathbf{drug} \mathbf{at} \mathbf{concentration} \mathbf{x}}$$

Where NCN = Normalized colony number relative to control;

NCN_2Gy = average colony number of 2 Gy only / average colony number of non-IR well (“Ctrl”);

NCN_{chemo+drug at concentration x} = average colony number of chemotherapy + drug only at concentration x / Ctrl;

NCN_{chemo+2Gy+drug at concentration x} = average colony number of chemotherapy + drug at concentration x + 2 Gy / Ctrl.

The downstream principal components analysis was done and visualized by factoextra R package. The heatmaps were generated by ggplot2 R package. And the clonogenic survival curve was generated by CFAssay R package [16]. All the data processing and visualization were done in R. We generated a 2-step process to determine whether a tested drug has potent CRT sensitization effect. First we visualize the individual drugs based on their SRF values at different test conditions on 2-dimensional space and the CRT sensitizers should be well-separated from the negative controls like DMSO. Second, a stringent criterion was applied to determine a drug’s CRT sensitization effect: only the tested drug have SRF value equal or higher than 1.5 at at least 4 out of the 6 tested drug concentrations can be identified as potent CRT sensitizer for the corresponding cell line.

Results

Integration of the high-throughput CSA assay within the framework of the U01 chemoradiation sensitizers screening platform

Appreciating the opportunities and the challenges of our efforts to find CRT sensitizers that are more efficiently translated from the bench to the bedside, we utilized the high-throughput CSA assay to screen the entire portfolio of CTEP agents in rapid, unbiased, uniformed, rigorous and carefully designed and executed preclinical studies. This was enmeshed within the larger ensemble of studies planned within our U01 grant portfolio, a three-pronged strategy to enable comprehensive investigation of CRT sensitization. The three-step strategy that will be employed uniformly across all molecular targeted agents in the CTEP portfolio. First, is the high-throughput screen for replicative cell death followed by in vitro standard clonogenic assay validation; second, the in vivo evaluation in multiple representative animal models; and third, mechanistic studies to delineate the molecular mechanism of the agent activity using multi-omics analysis (Figure 1A). While step 2 and step 3 are beyond of the scope of this paper, we highlight the optimization and reliability of the high-throughput screening method since this provides the key foundations for step 2 and step 3.

Figure 1.

Incorporation of the screening method in the U01 program. A. Overview of the U01 program. A 3-step strategy is deployed to rigorously test the CRT sensitization effect of a target drug in preclinical models. The high-throughput coonogenic survival screening method was incorporated in Step 1 and serve as key foundation for the whole procedure. B. Histogram showing the colony growth speed of a panel of 45 NSCLC cell lines. Each bar is color-coded by each cell line’s colony formation score (between 1–5) where 1 stands for the worst and 5 stands for the best. C. Cell growth curve for four selected NSCLC cell lines showing their intrinsic radiosensitivity when seeded at either 100 cells/well or 200 cells/well.

Optimization of the high-throughput screening method

Several key parameters need to be optimized to produce reliable and repeatable data using our in vitro high throughput screening method. Among them, the most basic but important parameter is the cell line selection. First, the ability of cell lines to grow into distinguishable colonies is one of the key aspects for a successful CSA experiment. Second, because of the limited number of cells that can be added into each well of the 96-well plates, the ideal number of days the cell lines need to grow into enough number of countable colonies (> 50 cells) will be less than 10 days. Therefore, to determine the feasibility of using each cell line in screening experiments, we tested the colony growth speed in a panel of 45 NSCLC cell lines in 96-well plate format. Moreover, we manually checked the colonies formed by each cell line and assigned a colony formation score to them based on key criteria on colony isolation, colony circularity, and average colony size. As shown in Figure 1B, about half of the tested cell lines were able to grow into countable colonies in less than 7 days and about one third achieved a colony formation score higher than 3. In order to get reliable data from the screening experiments, 13 cell lines (H460, A549, H1437, HCC15, H2087, H1944, H727, H1573, EKVX, HCC461, H358, H520 and HCC95) harboring representative mutations (i.e. TP53, KRAS) in NSCLC with colony growth speed less than 10 days and with colony formation score equal or larger than 4 were considered appropriate for use for the CRT sensitizer screening study.

Another important aspect that needs to be tested is the optimal initial number of cells seeded into the wells of 96-well plate. It will be difficult for the cells to form enough number of colonies when the initial seeding density is sparse while the over-growth of cells when the initial seeding density is too abundant is also problematic. Therefore, to determine the optimal initial seeding density, we compared the radiosensitivity of a panel of IR-treated NSCLC cell lines with 100 cells/well initial seeding density and 200 cells/well initial seeding density (Figure 1C, Supplementary Figure 1). Most of the tested cell lines tend to be more resistant to radiotherapy with higher initial cell seeding density. Moreover, the intrinsic radio-sensitivity of a cell line significantly affects its optimal initial seeding density and determines its eligibility to be used for screening experiments. Since 2Gy is the most clinically-relevant dose for radiation, cell lines like H358 with survival fraction less than 0.5 at 2 Gy radiation dose would not be appropriate for screening studies. For cell lines like EKVX and H1437 that are more resistant to radiation, a smaller initial seeding density (100 cells/well) should be used. And for cell lines like A549 that are more sensitive to radiation, a higher initial seeding density (200 cells/well) should be used.

Next, we sought to determine whether the screening system could identify some known radiation and chemoradiation sensitizers such as MEKi and HDACi that we reported before [15, 17]. We picked two less radio-sensitive cell lines (H460 and H1437) harboring either a Kras mutation (Q61H) or TP53 mutation (P278S), respectively, and one more radio-sensitive cell line (H358) harboring Kras (G12C) and TP53 null mutations, for the test experiment in 96-well plate format. The three cell lines were treated with radiation and 12.5 nM Trametinib (MEKi) or 1μM Vorinostat (HDACi) (Figure 2A and Figure 2B). Consistent with what we reported before, both Trametinib and Vorinostat are identified as potent radio-sensitizers in H460 (DER>1.5) by our screening method. Interestingly, we did not see significant sensitization effect of both drugs in H1437, which is probably due to the different genetic background between H460 (KRAS mutant, TP53 wild-type) and H1437 (TP53 mutant, KRAS wild-type). Another important observation is that although H358 also harbors KRAS mutation, we cannot identify the sensitization effect of trametinib because the cell line is too sensitive to radiation and drug so that there are very few colonies left after treatment. Therefore, it is also very important to consider the intrinsic radiosensitivity and the genetic background of the cell lines when incorporating them into screening studies.

Figure 2.

Optimization of the screening method. Clonogenic survival assay conducted in 96-well plate format with three NSCLC cell lines treated with combination of radiation and 12.5 nM Trametinib (Panel A) and 1 μM Vorinostat (Panel B).

Overview of the high-throughput CRT sensitizers screening method

The workflow of our high-throughput CRT sensitizers screening method as illustrated in Figure 3 consists of five major steps: 1) initial cell seeding on day 1; 2) treating cells with drug (chemotherapy + target therapy) and radiation on day 2; 3) stain the cells with crystal violet and count colonies with more than 50 cells using high-content imaging analysis between day 7 and day 10; 4) process the raw data and calculate the short-term radiosensitization factor at 2 Gy + chemotherapy (SRF_2Gy+Chemo); 5) downstream analysis for data visualization (i.e. principal component analysis, heatmaps, etc.) and identifications of CRT sensitizers. In step 1, to minimize the effects of medium evaporation to cell growth, we recommend users only seed cells into the inner 60 wells (grey wells highlighted in Figure 3A) and add PBS into the rest of the wells. Moreover, to avoid differences caused by the well position in the 96-well plate, we recommend users to seed cells with the same treatment conditions into wells at the same well position from different 96-well plates for biological triplicates. Another important note is to plan seeding an extra set of control cells which can be stained quickly to determine whether the cells have grown into desired number of colonies yet. In step 2 (Figure 3B), 96-well drug plates should be prepared in advance so that cells can be treated at the desired drug concentration. DMSO should be used to treat control cells, and the DMSO concentration should be consistent across all the conditions at any drug concentration. In step 3 (Figure 3C), we previously reported the detailed high content imaging analysis protocol [15]. An important note is to make sure the stained plates are completely dried before sending for high-content imaging analysis. In step 4 (Figure 3D–E), to determine the CRT sensitization effect of the tested drugs, we adapted from the concept of the SRF_2Gy [14] to calculate the SRF_2Gy+Chemo value at each drug concentration. Users should keep in mind that only colonies with more than 50 cells should be counted and included into the calculation. In step 5 (Figure 3F), we find dimensionality reduction analysis like principal component analysis (PCA) and plotting the SRF values at various drug concentrations with heatmaps are very helpful for data visualization and CRT sensitizer identification when there are large number of drugs screened within the same experiment. In PCA analysis, the potential targets usually tend to be clustered together in the two-dimensional space and well-separated from DMSO and other targets which do not have significant CRT sensitization effect. We recommend users to use more stringent criteria to identify potential CRT sensitizers when multiple drug concentrations are tested. We only count a drug as CRT sensitizer when it has SRF_2Gy+Chemo value equal to or higher than 1.5 at at least 2/3 of the tested drug concentrations [14].

Figure 3.

Workflow of the high-content clonogenic survival screen to identify CRT sensitizers. A. Cells were seeded in the inner 60 wells (highlighted) of 96-well plates with triplicates at the same well position at separate 96-well plates on day 1. B. Combination of radiotherapy and drug treatment were delivered on day 2. C. Cells were stained with crystal violet after 7–10 days incubation and automatically enumerated by IN Cell Analyzer 6000. D. Raw data were generated and only colonies with 50 cells or more were included in the downstream analysis. E. Formula for SRF_2Gy+Chemo value calculation at each tested conditions. F. Downstream analysis to visualize the screening results and identified potential CRT sensitizers (Drug 1- Drug 5).

Screening experiment identified Trametinib (MEKi) as a potent CRT sensitizer

After optimizing the screening system and showing that the method was able to produce consistent data as performed in conventional CSA experiments, we did a screening study to identify CRT sensitizers from 45 CTEP drugs including Trametinib as positive control using three NSCLC cell lines H460, H1437, and H727 that harbor different genetic background (Supplementary Figure 2C). First, since the cells need to be treated with both radiation and targeted drugs in the same day, we sought to test whether the sequence of treatment could affect the final results. Principal component analysis showed that different treatment sequences are mostly overlapped with each other (Figure 4A), suggesting that the treatment sequences are not significantly affecting the screening final results. Next, we tested the CRT sensitization effect of the 45 CTEP drugs at 6 different concentrations (0.1 nM, 1 nM, 10 nM, 100 nM, 1 μM, and 10 μM) in combination with the most clinically relevant chemotherapy dose (1.5 ng/ml Paclitaxel and 4.5 ng/ml Carboplatin) and radiation dose (2 Gy) that we’ve been consistently using for NSCLC in vitro experiments [Yifan 2018, Yifan 2018]. From the PCA result (Figure 4B) for each individual tested drug, it clearly showed that the positive control Trametinib together with some other screening “hits” are well-separated from the negative control DMSO and some other drugs that do not have significant CRT sensitization effect, including the IDO1 inhibitor (INCB-024360) which we don’t expect it to have any effect in in vitro models because of the lack of immune system. Next, we further calculated the SRF_2Gy+Chemo value for each tested drug at each concentration in the three NSCLC cell lines and visualized the result in heatmaps (Figure 4C). Consistent with our previous findings, Trametinib only has potent CRT sensitization effect in H460 which only harbors KRAS mutation but with wild-type TP53 genotypes. We further validated Trametinib’s CRT sensitization effect in two NSCLC cell lines (H460 and H727) and two PDAC cell lines (Panc-1 and L3.6pl) in conventional CSA experiments (Supplementary Figure 2B). Trametinib’s CRT sensitization effect was validated in both cancer types. Overall, we demonstrated that our high-throughput method is able to reliably identify potent CRT sensitizers.

Figure 4.

High-content clonogenic survival screen using 3 NSCLC cancer cell lines (H460, H1437 and H727) identified Trametinib as potent CRT sensitizers. A. Principal component analysis for experiments done with six different sequences of treatment combination of CRT and targeted drugs. B. Principal component analysis for individual drugs from the screening results. The positive control Trametinib and negative control DMSO and INCB 024360 were labeled on the plot. C. Heatmaps visualizing the SRF_2Gy+Chemo value at each tested drug concentrations for the three cell lines. The positive control Trametinib and negative control DMSO and INCB 024360 were labeled.

Discussion

In the present study, we demonstrated that our high-throughput screening method in 96-well plate format is able to reliably identify potent CRT sensitizers. This is the first study to systematically assess the feasibility of using 96-well plates to identify CRT sensitizers. We described the whole procedures to optimize and perform the screening method, with details on determination of cell line growth speed, colony formation score, and initial cell-seeding density, consideration of the genetic background and intrinsic radiosensitivity of cell lines, treatment sequences, and combination with CRT as well as RT. Here, we summarize the key parameters that we think are vital for a successful screening experiment in Table 1. Briefly, first, only the cell lines that can grow into well-isolated colonies within 10 days in 96-well plates should be considered to be used for the screening experiment. Second, we showed that the intrinsic radio-sensitivity of a cell line could profoundly affect the screening results. Hence, cell lines that are very sensitive to radiation should be excluded for screening experiments. For NSCLC cell lines, we previously have reported the intrinsic radio-sensitivity in a panel of cell lines [18], which can serve as a reference for cell line selection. Third, it is very common that certain drugs will only sensitize cell lines with specific genetic mutations. Therefore, it is crucial to include multiple cell lines that harbor different mutations in important genes that are related to certain cancer types. Fourth, it is very important to include both negative control and positive control in the screening experiment. Date generated from positive control and negative control can be very helpful for screening experimental quality assessment. Lastly, an important observation we found was treatment sequences did not significantly affect the final screening results, suggesting the screening results are comparable as long as the treatment sequences are consistent. However, we still recommend to keep the gap between radiotherapy and drug treatment in less than four hours.

Table 1.

Checklist for key parameters.

Procedures	Key parameters
Cell Line Selection	1. Cell Colony Growth Speed 2. Cell Colony Formation Score 3. Intrinsic Radiosensitivity and Drug sensitivity 4. Genomic Background
Screening Set-ups	1. Initial cell seeding density 2. Triplicates should be planned at the same well position from separate 96-well plates 3. Positive and negative controls should always be included 4. Drug plates should be prepared in advance, and DMSO concentration should be the same across all drug conditions 5. Plan extra wells for negative control, which can be stained in adavnce to determine whether the cell colonies are forming as expected 6. Only use the inner 60 wells for cell seeding and add equal volume of PBS into the outter 36 wells to minimize the evaporation effect on cell growth 7. Cell plates should be put back to incubator as soon as the treatment is done and maintaied undisturbed until ready for staining
Data Processing & Analysis	1. Make sure the stained cell plates are completely dried before sending for high-content imaging analysis 2. Always calculate the standard deviation of the triplicates to determine whether the data is usable for down-stream analysis 3. Use stringent criteria to determine CRT sensitizers
Validation	1. Validate the identified hits by conventional CSA in vitro 2. Validate the identified hits by immune-compromised models in vivo 3. Validate the identified hits by immune-competent models in vivo

Since our method will significantly reduce the amount of time required for CRT sensitizers identification comparing to conventional clonogenic assays, our screening system can serve as a reliable platform to facilitate the discoveries of potential actionable targets that increases the clinical efficacy of CRT. However, due to the nuances and complexities of human physiology and cancer biology, before translating into clinical use, all the identified CRT sensitizers identified from our method should be validated vigorously in preclinical settings, including traditional CSA in vitro validation, in vivo validation using immune-compromised models for tumor-intrinsic sensitization factors, and in vivo validation using immune-competent models for tumor microenvironmental sensitization factors. Moreover, radiation dose standardization also plays important roles in guarantee the reproducibility of in vitro data, which we and others [5–8, 19, 20] have published guidelines elsewhere.

While our method allows for simultaneously screening for CRT sensitizers among hundreds of drugs, it would be very laborious when larger scale of drug screening studies are needed. Since our screening method uses colony-formation as read-outs, we are limited to adapt to higher throughput format, 384-well plate for example, because of the need to have enough number of colonies to count for. Therefore, development of surrogates of colony formation that can stably reflect the long-term survival of cells is greatly needed. Another potential weakness of our method is that it is only applicable to cell lines that are able to grow into distinguishable colonies. Therefore, for certain cell lines that form poor colony morphologies, our screening method might not be able to generate high quality data. However, for most of the time, we will be able to find surrogate cell lines with similar genetic background that can form well-distinguishable colonies.

Moreover, with the rapid development and reduced cost of next-generation sequencing technology, genomic, transcriptomic, epigenetic, proteomic and metabolomic profiling studies have been more involved in preclinical studies to elucidate the molecular mechanisms for identified radiation/CRT sensitizers. We believe by incorporating multi-omics analysis into the screening system, it will further facilitate the development of CRT sensitizers in the era of precision radiation medicine. And with the advent of single-cell genomic technology, it becomes feasible to use genomic data as read-outs for large-scale screening studies. A recent study showed the ability to use multiplexed single-nucleus transcriptomic data as read-outs for high-throughput screening of around 200 chemicals [21]. Another recent study showed the feasibility of co-culturing different cell lines without the need to barcode each cell line within the same well/plate and using single-cell RNA-seq data as read-outs for cell line identification [22]. This method potentially would significantly increase the throughput of the current CRT sensitizers screening method.

In general, our high throughput screening method facilitates rapid and accurate identification of the most potent CRT sensitizers in a format that allows testing dosing and sequencing strategies seamlessly, quantitatively, reproducibly and rapidly. The future incorporation of multi-omics analysis on both single-cell and bulk-tissue level would greatly facilitate the development of precision radiation/chemoradiation therapy.