Phenotypical screening on metastatic PRCC-TFE3 fusion translocation renal cell carcinoma organoids reveals potential therapeutic agents

All experiments with human tissue were approved by the Ethics Committee of the National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences (NCC/CHCAMS) (ID: 20/246-2442), and written informed consent was obtained. Renal tumor tissue was obtained from treatment-naïve patients who underwent nephrectomy at the Department of Urology, NCC/CHCAMS. The sample was processed within 24 h after surgery. Moreover, pathological evaluation and TFE3 break-apart fluorescence in situ hybridization (FISH) detection confirmed the malignancy of the samples and the TFE3 fusion status.

On arrival, tumor tissue was extensively washed with cold phosphate buffered saline (PBS) and then minced into cubes smaller than 1 mm³ on ice. Subsequently, the minced tissue pieces were digested with collagenase (2 mg/ml, Cat. No. C9407, Sigma–Aldrich, St Louis, MO, USA) for 1 h at 37 °C and filtered through sieves with 100-μm pores. After the samples were washed twice with fresh medium (2% fetal calf serum, FCS; advanced DMEM/F 12, 2 mM HEPES, 1 × GlutaMAX-I, 200 U/ml penicillin/streptomycin) and centrifugation (300 g, 5 min), the dissociated cells were seeded into growth factor-reduced Matrigel (Corning Inc., Corning, NY, USA) in the presence of advanced DMEM/F12 at 37 °C for 30 min. Next, the surface of solidified mixture of cell suspension/Matrigel was sealed with complete human organoid medium (HOM, 1 ml), which was comprised of Advanced DMEM/F12 supplementing with series additives (1 × B27 supplement, 10 nM Lue15-Gastrin I, 1 mM N-acetylcysteine, 100 ng/ml recombinant human IGF-1, 50 ng/ml recombinant FGF-2, 20% afamin/Wnt3a conditional medium, 1 µg/ml human R-spondin, 100 ng/ml Noggin, 500 nM A-8301, 200 U/ml penicillin/streptomycin, and 10 µM Y-27632) as described by Lampis et al. [11] and Loredana et al. [12]. The medium was replaced every 3 days when the diameter of organoids ranged up to 200 ~ 300 μm, and the organoids were dissociated and passaged weekly. Matrigel was broken up by pipetting up and down several times, and organoids were collected in a tube. After centrifugation at 300 g for 5 min, organoids were dissociated using TrypLE Express (Gibco, Grand Island, NY, USA) for 10 min. Cell clusters were reseeded at a typical split ratio of 1:3. A bright-light microscope (LEICA, Wetzlar, Germany) was used for organoid bright field photographing. Early passage organoids were frozen in Recovery Cell Culture Freezing Medium (Gibco) and stored at − 80 °C before drug screening. The whole process from tissue collection to compound screening is shown in Fig. 1A. A detailed protocol will be provided by the authors upon request.

Tumor tissues and organoids were fixed in 4% paraformaldehyde followed by dehydration, paraffin embedding, sectioning and standard hematoxylin and eosin (H&E) staining. Immunohistochemistry (IHC) staining was performed with TFE3 (dilution-free. RMA-0663, Shandong Maixin Biotechnology Co., China), E-cadherin (dilution 1: 200. MAB-0738, Shandong Maixin Biotechnology Co., China), and Ki-67 (dilution 1: 200. GM7240, GeneTech Co., Ltd., Shanghai, China) antibodies on an Autostainer 480S (Fa Medac). Images of stained samples were taken with an Olympus DP73 microscope (Olympus, Japan).

TFE3 protein expression confirmed by IHC is routinely used for preliminary diagnosis, but clinically FISH is the gold standard for diagnosis of TFE3 fusion tRCC [13]. A dual-color, break-apart FISH assay was performed to detect TFE3 using the TFE3 (Xp11.2) break probe set (Guangzhou LBP Medical Technology Co., Ltd., China). FISH signals were assessed under an Olympus BX51TRF microscope (Olympus, Japan). Normal cells without fusion had one (for males) or two (for females) yellow signals. Signals were considered to be split when the distance between red and green signals was ≥ 2 signal diameters. For each case, a minimum of 100 tumor nuclei were evaluated. Overlapping cells indistinguishable as separate nuclei were not included in the analysis to avoid false positives due to nuclear truncation. A positive result was defined when ≥ 15% of the tumor nuclei had split signals.

HTS was performed automatically using a mechanical liquid handling equipment mechanical arm (NAYO N96-204, NAYO Biotech. Co., Ltd. Shanghai, China). A drug library containing 1816 small molecule compounds was obtained from MedChemExpress (Shanghai, China, catalog number HY-L0022M) with modification (Supplementary Data 1). The library was reformatted into 96-well source plates with a concentration of 1 mM for automated robotic screening. In parallel, the cells were also treated with an equal volume (0.2%) of dimethyl sulfoxide (DMSO) as a negative control and 2.5 μM final BEZ-235 (MCE, Shanghai, China) as a positive control. BEZ-235 is an inhibitor of phosphatidylinositol 3-kinases (PI3-K) and is commonly used as a positive control in HTS or drug exploration [14, 15]. Plate-to-plate normalization and assay quality control were performed according to them. A 3D cell viability assay was implemented to determine the number of viable cells according to ATP level using commercially available luminescence detection reagent (CellTiter-Glo #G9683, Promega, Madison, WI). Briefly, organoids were processed as described previously [16] and plated in a 96-well low binding assay plate at a density of 8,000 cells per well in 50 μl 10% growth factor reduced Matrigel. Additional 40 μl culture medium without Matrigel was added as described above. Organoids were maintained in medium described earlier [17] and treated 2 days later by adding 10 μl culture medium with 10 μM compound to obtain a final concentration of 1 μM. The assay was terminated at day 5 by adding 50 μl CellTiter-Glo. Assay quality and robustness were evaluated with the signal window (SW) and Z factor. The SW and the Z factor have been routinely used in previous HTS studies for validation [18‐20]. Triplicate wells treated with BEZ-235 and vehicle solvent (DMSO) were employed as bottom wells and top wells, respectively. When SW was above 6 and the Z factor values were between 0.44 and 0.78, the assay was qualified for HTS. The equipment of the Z factor is listed below (σ, standard deviation. μ, average. p, positive control. n, negative control) [21] as follows:

Cell cycle and apoptosis were detected as previously described [22]. Organoids were cultured and treated with the indicated drugs at different concentrations for 48 h, followed by single staining with propidium iodide (Beyotime Biotechnology Co., Ltd., China) for cell cycle analyses and double staining with propidium iodide and Annexin V-FITC (Beyotime Biotechnology Co., Ltd., China) for apoptosis analyses.

Organoids were incubated with DMSO or the indicated drugs for 120 h. After harvesting, total RNA was extracted using the TraZolTM UP Plus RNA Kit. RNA was sent for sequencing and analysis (BGI Co., Ltd., Shenzhen, China). Briefly, total RNA was fragmented, and mRNA was enriched using oligo (dT) magnetic beads, followed by cDNA synthesis. Double-stranded cDNA was purified and enriched by PCR amplification, after which the library products were sequenced using BGIseq-500. The volcano plot and heatmap of differentially expressed genes (log2 of fold change > 1, q < 0.05) were generated by BGI using the Dr. TOM approach, a customized data mining system from BGI. Next, 232 autophagy-related genes (ARGs) were obtained from the Human Autophagy database (HADb, http://​www.​autophagy.​lu/​index.​html) and used for analyses.

Data statistical analysis was performed using Prism 6.4. The half-maximal inhibitory concentration (IC₅₀) values were analyzed using nonlinear regression (curve fit). The 95% confidence interval was calculated. The differences in IC₅₀ values between organoids were analyzed using the nonparametric Mann–Whitney U test. Cell cycle and apoptosis data were analyzed using Excel with Student’s t test. q < 0.05 was considered to indicate statistical significance.

One 21-year-old male patient was diagnosed with right kidney cancer and retroperitoneal lymph node metastasis. Magnetic resonance imaging (MRI) indicated a solid tumor of 6.0 × 5.0 × 5.2 cm located in the lower pool of the right kidney with arterial phase enhancement (Fig. 1B). Laparoscopic nephrectomy and retroperitoneal lymph node dissection were performed in April 2019. The pathology indicated RCC and lymph node metastasis (T1bN0M1, Stage IV) (Fig. 1C). H&E staining demonstrated that the tumor featured papillary architecture combined with foci of necrosis and gravel-like calcification (Fig. 1D). IHC showed positivity for Ki-67 and TFE3 protein expression (Fig. 1E and F). tRCC was confirmed by the FISH assay (Fig. 1G).

After obtaining freshly resected tumor tissue, we successfully established patient-derived tRCC organoids (Fig. 2A and B). The established tRCC organoids were propagated for three or more passages and cryopreserved. H&E data showed that both organoids and original tissue displayed a cytoplasm papillary structure harboring multiple lumen epithelium folds and invaginations (Fig. 2C). These data demonstrated the association between the morphological structure of tRCC tumors and their derived organoids. IHC staining of Ki-67 (Fig. 2D) and E-cadherin (Fig. 2E) indicated that organoids retained epithelial origin and proliferation activity.

TFE3 break-apart FISH is currently the gold standard for the identification of TFE3 rearrangements [13]. TFE3 break-apart FISH of the derived organoids indicated that the cells with split signals accounted for 69.5% of the signals (Fig. 2F). Furthermore, RNA-seq was performed, and a fusion of PRCC (exon 1, end with protein IAAPELHKGD) and TFE3 (exon 6, starting with protein CLCQGICLMC) was detected with a frequency of 15% (Fig. 2G).

Our results indicated that the tissue architecture, phenotype, cellular composition, and typical PRCC-TFE3 gene fusion of the primary tumors were reliably conserved in PRCC-TFE3 fusion organoid models.

After successfully establishing and characterizing PRCC-TFE3 fusion tRCC organoids, we focused on developing a three-dimensional (3D) cell-based assay that was suitable for HTS. HTS requires robust assays with high reproducibility of well-to-well and plate-to-plate assays, as well as high signal-to-noise ratios. To determine whether PRCC-TFE3 organoids could be seeded and cultured in 3D, we established a workflow employing an automated liquid-handling system (Fig. 3A).

Then, we focused on the validation of assay robustness and reproducibility by assessing plate uniformity using the parameters of SW and Z factor. Three independent experiments were conducted to evaluate the day-to-day variation. In each experiment, 10 plates were used to measure plate-to-plate variation. High control and low control wells were defined as organoids treated with medium containing DMSO and BEZ-235. Each treatment well was analyzed in quadruplicate to determine the well-to-well variation. The Z factors ranged from 0.44 to 0.78, indicating the signal-to-noise reproducibility between plates was high enough. Z factors of three independent experiments were 0.61, 0.61, and 0.66 showing the day-to-day variation was low. The average values of for SW and the Z factor were 6.88 and 0.63, respectively (Fig. 3B). These results indicated that the assay was qualified for HTS with high reproducibility. Representative images of organoids at different time points are shown in Fig. 3C–J.

To facilitate the screening of thousands of compounds across hundreds of cell lines, we used a small-molecule library containing 1,816 compounds. The target average final concentration of the compound library for screening in cell-based assays was 2 µM. Most compounds (1271, 70%) were non-oncology-related, and the remaining compounds were either chemotherapeutics or targeted oncology agents.

We designed a 2-stage screening strategy whereby the 1816 drugs were primarily screened in a single well at a single dose (2 μM). A total of 101 drugs showed more than 50% inhibition (Fig. 4A). On the other hand, when analyzing the distribution of activity with the compound category, most compounds were related to epigenetics, tyrosine kinases, the cell cycle and autophagy (Fig. 4B). Interestingly, among the 101 active drugs, most active compounds (61 out of 101, 60.4%) were originally developed for non-oncology clinical indications (Supplementary Fig s1).

As we aimed to use phenotypical drug screening to identify medicines for repurposing and assess potential mechanisms of action for TFE3 fusion tRCC, 45 chemotherapy agents with nonspecific cytotoxicity were excluded from the 101 compounds. After that, 56 compounds were retained, and axitinib was manually added to the panel. Eventually, a drug panel containing 57 compounds was designed (Supplementary Data 2) and subjected to testing via an eight-point dose–response model (doses ranging from 60 μM to 27 nM). Finally, three representative drugs, axitinib, crizotinib, and JQ-1, were selected, and the dose–response results indicated that all of them significantly inhibited cell viability (Fig. 4C).

To be noticed, axitinib is a small molecule tyrosine kinase inhibitor that inhibits multiple targets, including vascular endothelial growth factor receptor. As the PDOs did not contain vascular endothelial cells, axitinib only killed 35.4% of cell in PRCC-TFE3 fusion tRCC organoids. Axitinib activity in RCC patients has been studied in various settings, particularly as a second-line targeted treatment. Although axitinib did not reach 50% inhibition, we also included it for further investigation.

As TFE3 factors in tRCC can regulate autophagy, we then focused on autophagy-related drugs. Autophagy is required for crizotinib-induced apoptosis in MET-amplified gastric cancer cells [23]. JQ-1 is a BRD4 inhibitor, and BRD4 is a transcriptional repressor of autophagy and lysosomal function [24]. Therefore, axitinib, crizotinib, and JQ-1 were selected for further investigation.

To further explore the cellular mechanism of axitinib, crizotinib, and JQ-1 in PDOs, we investigated the cell cycle and apoptosis effects. Proliferation assays demonstrated that all three drugs could inhibit cell proliferation in dose-dependent and time-dependent manners (Fig. 5A–C). The bright-field images (200 μm) of tRCC PDOs after incubation of 120 h with DMSO, axitinib, crizotinib, and JQ-1 showed that compared with the vehicle control (DMSO), three drugs of low-dose or high-dose induced the cell apoptosis significantly (Fig. 5D–F). Fluorescence-activated cell sorting (FACS) showed that axitinib, crizotinib, and JQ-1 induced changes in cell counts during the cell cycle (Fig. 5G). After quantitative analyses, the percentage of cells during the cell cycle showed that JQ-1 induced G0/G1 phase arrest, while axitinib and crizotinib induced G2/M phase arrest (Fig. 5H). All three drugs promoted cell apoptosis, especially early apoptosis, in a dose-dependent manner (Fig. 5I), and the cell apoptosis rates with these three drugs were shown in Fig. 5J.

To explore the underlying molecular mechanism of the antitumor effect of axitinib, crizotinib, and JQ-1, we performed whole-genome mRNA expression profiling. Total RNA isolated from PDOs treated with axitinib, crizotinib, and JQ-1 was analyzed by 3’ mRNA-Seq and compared to that from PDOs treated with DMSO. Using a log2 of fold change cutoff of 1 and a q-value cutoff of 0.05, we identified differentially expressed genes. In the axitinib high-dose group, only ten genes were upregulated, and one gene was downregulated. In the JQ-1 high-dose group, eight genes were upregulated, and 28 genes were downregulated. In the crizotinib high-dose group, 364 genes were upregulated, and 671 genes were downregulated (q < 0.05, log2 of fold change > 1) (Fig. 6A) (Supplementary Data 3).

We observed that ARGs were modulated by JQ-1 and crizotinib but not by axitinib (Fig. 6B). In particular, crizotinib treatment showed significant effects on ARGs (Fig. 6C) (Supplementary Data 4).