Liquid-based 3D Cultured Tumoroids Modeling Resistance to Cisplatin and Imatinib in Metastatic Colorectal Cancer

Researchers have developed and used several three-dimensional (3D) culture systems, including spheroids, organoids, and tumoroids. Drug resistance is a crucial issue involving recurrence in cancer patients. Many studies on anticancer drugs have been done in 2D culture systems, whereas 3D cultured tumoroids have many advantages for assessing drug sensitivity and resistance. Here, we aim to investigate whether Cisplatin (a DNA crosslinker), Imatinib (a multiple tyrosine kinase inhibitor), and 5-Fluorouracil (5-FU: an antimetabolite) alter tumoroid growth of metastatic colorectal cancer (mCRC). To establish a liquid-based 3D multiplexing reporter assay system, LuM1 (a murine mCRC cell line) was stably transfected with the Mmp9 promoter-driven ZsGreen reporter gene, which was designated as LuM1/m9 cells and cultured in NanoCulture Plate (NCP), a 3D culture device. The larger tumoroids were not sensitive to Cisplatin and expressed ABCG2 (a marker of cancer stem cells, a.k.a. a drug efflux transporter), whereas smaller cell-aggregates were more sensitive to Cisplatin. Both Imatinib and Cisplatin significantly increased tumoroid growth (larger than 300 μm2) and Mmp9 promoter activity and were not cytotoxic to the mCRC tumoroids. On the other hand, 5-FU was cytotoxic to the tumoroids and significantly inhibited tumoroid growth, although not completely. Thus, platinum resistance and imatinib resistance in mCRC were modeled using the liquid-based 3D cultured tumoroid system. The tumoroid culture is useful and easily accessible for the assessment of drug sensitivity and resistance.

Drug resistance is a crucial issue in cancer, which occurs recurrence or relapse in cancer patients. Some chemoresistance has been overcome; however, recent studies clarified resistance against molecularly targeted therapeutics, including antibody drugs [25,26]. Many studies on drug resistance have been done in 2D culture systems, whereas we recently have examined chemosensitivity and chemoresistance in the liquid-based 3D cultured tumoroids [13,14]. For example, 5-FU, an antimetabolite inhibiting DNA replication, inhibited tumoroid growth of metastatic colorectal cancer (mCRC), while this agent did not inhibit metastatic metalloproteinase activity of mCRC cells [14]. On the other hand, cortisol, an anti-inflammatory steroid hormone agent, inhibited metalloproteinase activity of the mCRC, whereas this agent did not inhibit but rather promoted tumoroid growth [14]. Moreover, using the liquid-based 3D culture system, we have identified repurposing drugs for cancer therapy: (1) artesunate, an anti-malaria drug, inhibits tumoroid growth of mCRC [14]; (2) benztropine, an anti-Parkinson disease drug, inhibits tumor growth and metastasis of mCRC [13].
However, chemosensitivity and chemoresistance are not fully investigated in 3D culture systems. DNA crosslinkers such as cisplatin and tyrosine kinase inhibitors (TKI) such as Imatinib (Gleevec) have not been well examined in 3D tumoroid culture systems. It is also not well studied whether 5-FU, a cell-cycle-dependent antimetabolite, is effective to tumoroids that include slow cycling, dormant cancer cells. Thus, in the present study, we aim to investigate whether Cisplatin, Imatinib, and 5-FU alter tumoroid growth of metastatic cancer cells in the liquid-based 3D culture system, which can model a new system for analysis of chemoresistance and chemosensitivity.

Cells and 3D culture
A murine colorectal cancer cell line Colon26 and its rapidly metastatic subline LuM1 were used [22,27,28]. LuM1/m9 reporter cells were established by the stable transfection of a murine Mmp9 promoter (588 bp)-driven ZsGreen reporter construct into LuM1 cells [13,14]. These cell lines were maintained in RPMI 1640 containing 10% FBS supplemented with penicillin, streptomycin, and amphotericin B.

Drug treatment
For analysis of in vitro tumorigenesis and chemoresistance, cells were seeded at a concentration of 5 × 10 3 or 1 × 10 4 cells per well in a 96-well NCP and then pre-cultured in mTeSR1 for 1, 3, or 4 days. Cisplatin was added to the pre-formed tumoroids at 10 μM, tumoroids were cultured 24 hours before viability assay. LuM1/m9 cells were seeded at 5 × 10 3 cells/well in a 96-well NCP and then cultured in 10% FBS-containing RPMI 1640 medium for 72 h with or without drugs (Cisplatin, Imatinib, or 5-FU) in the other experiments. After each drug's treatment, the size and viability of cell aggregate/tumoroid were measured as described below.

3D Tumoroid-based multiplex reporter assay
Tumoroids of LuM1/m9 cells were formed in NCP as described above and previously [7,13,14]. The fluorescent area of each tumoroids or cell aggregates were analyzed using the ArrayScan HCS System (ThermoFisher Scientific, Waltham, MA). Fluorescent areas greater than 300 μm 2 were considered as tumoroids. Mmp9 promoter activity was evaluated by an average fluorescence intensity per μm 2 of all cells in a well. Experiments were performed with 3 or 4 biological replicates.

Cell viability assay
The ATP content was quantified using a CTG luminescent cell viability assay (Promega, Medison, WI). Briefly, from the total 200 μl of culture supplement, 150 μl was removed, and 50 μl of CTG solution was added to each well and then suspended. The plate was rocked for 2 min and incubated for 10 min at 37ºC. The luminescence was measured in a plate reader (Molecular Devices, San Jose, CA).

Side population cell analysis
Cells cultured on a 10-cm dish were washed with warmed PBS (-), detached using 5 ml of Accutase (Innovative Cell Technologies, San Diego, CA), and neutralized with 5 ml of the serum-contained medium. Cells were collected by centrifugation at 250 × g for 3 min and suspended in the serum-contained medium. Cell aggregates were removed by using a 35-μm cell strainer. The density of cells was adjusted to 10 6 cells/ml by adding the serum-contained medium. Cells were incubated at 37ºC by using a water bath, mixed with Hoechst 33342 (ThermoFisher Scientific) at a final concentration of 5 μg/ml and/or verapamil (Sigma) at a final concentration of 30 μg/ml, and then incubated at 37ºC for 90 min in a dark condition with stirring every 20 min. Cells were then centrifuged at 250 × g for 3 min and washed with ice-cold PBS (-) containing 2% FBS. The cell density was adjusted to 10 7 cells/ml and mixed with Propidium Iodide (PI) at a final concentration of 0.5 to 1.0 μg/ml. Cell aggregates were removed using the cell strainer, and single cells were analyzed using a FACSCalibur (BD Biosciences, Franklin Lakes, NJ) at Central Research Laboratory, Okayama University Medical School.

RT-qPCR
RT-qPCR was performed as described [13,15]. LuM1 cells were cultured in 2D or 3D conditions for 4 days, and total RNA was extracted using the AGPC method with Trizol (Molecular Research Center, Cincinnati, OH). cDNA was synthesized using ReverTra Ace (Toyobo, Osaka, Japan). Real-time PCR was carried out using iQ cyber (BioRad). Primers for Abcg2 and Hprt1 (an internal control) were listed in the previous report [15]. Relative levels of Abcg1 mRNA to Hprt1 mRNA were quantified by the ΔΔCt method using the formula as follows: fold change = 2 −ΔΔCt .

Statistical analysis
Data were expressed as the means ± SD unless otherwise specified. Statistical significance was calculated using GraphPad Prism (La Jolla, CA). Three or more mean values were compared using one-way analysis of variance with the pairwise comparison by Turkey's method, while two were made with an unpaired student's t-test. P<0.05 was considered to indicate statistical significance.

Tumoroids acquired platinum-resistance with ABCG2 expression in metastatic colorectal cancer cells
We have shown that the mCRC cell line LuM1 expressed a high level of ABCG2, drug efflux pump expressed in CSCs, than the milder colorectal cancer cell line Colon26 [15]. To evaluate platinum resistance and cancer stem phenotype of the mCRC cell line LuM1, we first examined whether the rate of side population (SP) cells was reduced by verapamil, an ABCG2 inhibitor. The rate of SP cells in untreated LuM1 cells was 7.8%, while verapamil reduced it to 2.2%, suggested that this mCRC cell line contained ABCG2+ CSCs ( Figure 1A). The rate of SP cells in Colon26 was 2.9%, thus fewer than that of LuM1, while verapamil further reduced the SP cells in Colon26 down to 0.4% ( Figure S1).
Next, we examined whether the mCRC cells' platinum resistance was altered by 2D vs. 3D culture systems, the size of tumoroids, and the type of liquid environments (serum vs. stemness enhancing medium). The NCP-based 3D culture condition promoted tumoroid growth compared with a 2D culture system ( Figure 1B, C). Stemness enhancing medium (mTeSR1) also promoted tumoroid growth compared to the serum-contained medium. Cisplatin reduced cell viability (ATP activity) in 2D culture conditions and in the serum-contained medium. In contrast, the cytotoxicity of Cisplatin was reduced by larger tumoroid formation based on the 3D culture system and the stemness-enhancing medium. Notably, the combination of the 3D culture and mTeSR1 markedly promoted cisplatinresistant tumoroid growth ( Figure 1D). Moreover, the combination of the liquid-based 3D culture system and stemness-enhancing medium significantly increased Abcg2 gene expression.
These data indicated that 3D-cultured tumoroids with cancer stem phenotype were more platinum-resistant than 2D-cultured cancer cells. Larger tumoroids were more platinum-resistant than smaller cell aggregates. The 3D tumoroid formation of mCSC cells induced Abcg2 expression that may involve Cisplatin efflux.

Cisplatin promoted tumoroid formation of metastatic colorectal cancer
Next, we asked whether pre-formed, enlarged tumoroids were resistant to Cisplatin. Cisplatin did not alter viability and Mmp9 promoter activity in the mCRC LuM1 cells (Figure 2 A, B, C). Notably, Cisplatin tended to stimulate tumoroid growth ( Figure 2D, 2E). The large tumoroids (10,000 to 20,000 µm 2 ) were formed by cisplatin treatment, which was not found in the untreated group ( Figure 2F). Cisplatin treatment significantly increased the tumoroid size as compared with the untreated group ( Figure 2G). These data indicated that the Cisplatin is ineffective to pre-formed, enlarged tumoroids, promoting rather tumoroid growth.
Next, we examined a lower concentration ( large tumoroids > 300 µm 2 ( Figure 2K), and Mmp9 promoter activity ( Figure 2L). These data indicated that Cisplatin at 2 µM can promote tumor growth in mCRC cells.

Imatinib promoted tumoroid growth of mCRC
Next, we examined whether Imatinib, a tyrosine kinase inhibitor, altered the tumoroid growth of mCRC cells. Imatinib (10 µM) significantly promoted tumoroid growth ( Figure   3A, B, C). The rate of large tumoroids >300 µm 2 was significantly increased by imatinib treatment ( Figure 3B). The size of tumoroids was also significantly increased by imatinib treatment ( Figure 3C). The Mmp9 promoter activity was significantly increased by imatinib treatment. However, cellular viability was not altered by imatinib treatment (Figure 3E).
These data indicated that mCRC tumoroids of LuM1 were imatinib-resistant. Imatinib can promote cell aggregation and intercellular adhesion of mCRC cells.

Discussion
Drug resistance is an unsolved issue, especially in metastatic cancer, including mCRC. Many studies have examined chemoresistance in 2D culture systems, whereas our current study significantly demonstrated that the 3D culture system is a suitable, easily accessible tool to analyze drug resistance in metastatic cancer in vitro. Our study showed that Cisplatin (a platinum-based DNA crosslinker) and Imatinib (TKI) were ineffective and rather promote tumor growth of the metastatic CRC cells. On the other hand, 5-FU (an antimetabolite inhibiting DNA replication) more effectively combat the tumor growth of mCRC, although MMP9-positive metastatic cancer cells remained even after the 5-FU treatment, which was thought to be insensitive, dormant cancer cells in the tumoroids. Thus, the 3D tumoroid culture system is useful for further analyzing drug resistance and metastatic cancer cells' sensitivity.
Our study touch upon the mechanisms of platinum resistance using the liquid-based 3D tumoroid model. Tumoroid growth induces both ABCG2 expression ( Figure 1) and exosome secretion [6,7,15]. Therefore, it is conceivable that Cisplatin could be secreted from cancer cells with exosomes and via ABC transporters such as ABCG2 efflux pump ( Figure S2). Several studies showed that Cisplatin was secreted with exosomes from cancer cells [25,29,30]. Notably, we showed even anti-EGFR antibody-based drug (cetuximab) was secreted with exosomes from metastatic oral cancer cells [26]. Moreover, cancer cellderived EVs, including exosomes, promote tumor growth of mCRC in vitro and in vivo [6,21]. Therefore, Cisplatin-induced exosomes may promote tumor growth, which may be a mechanism by which Cisplatin promoted tumor growth.
Our data showed that Imatinib was not cytotoxic but promoted tumoroid growth of mCRC. Imatinib is a TKI that can inhibit multiple TKs, including receptor tyrosine kinases (RTK), in cancer cells. c-Kit is one of the RTKs targeted by Imatinib. Chau et al. reported c-Kit mediated chemoresistance and tumor-initiating capacity of ovarian cancer cells through activation of Wnt/β-catenin-ABCG2 signaling [31]. Moreover, Imatinib was reported to upregulate compensatory integrin signaling in a mouse model of the gastrointestinal stromal tumor (GIST) [32]. It was also shown that Src family kinase induced Imatinib resistance in a kinase-dependent manner in chronic myelogenous leukemia (CML) cells [33]. Src family kinase has been shown to mediate growth factor signals to multiple pathways, including STAT, Ras-MEK-ERK, and PI3K-Akt [32,34]. Among these pathways, STAT3 is crucial for activating the MMP9 gene and cancer stem phenotype in the 3D tumoroid model [13] ( Figure S2).
Moreover, Liu et al. reported exosome release in Imatinib-resistant human CML cells [35]. Also, it has been reported that ABCG2 transports anticancer agents such as Imatinib from cells [24]. Therefore, it was suggested that the integrins-Src-STAT3 signaling pathway, exosome release, and/or efflux via ABCG2 may cause Imatinib resistance. The only hope is that tumoroid formation is associated with the inhibition of epithelial-to-mesenchymal transition (EMT) [3]. EMT involves the migratory, invasive, and metastatic ability of cancer cells [12,36], and thus inhibition of EMT is a potent strategy for cancer therapy. Tumoroid formation with intercellular adhesion is a result of an anti-EMT effect of the drugs. Imatinib may inhibit EMT and thus inhibit the invasive, metastatic ability of cancer cells.
It has been known that many chemotherapeutics unselectively kill proliferating cells, including cancer cells and myeloid, in patients and thus often cause adverse effects, including myelosuppression. These limitations in chemotherapies have been overcome by molecularly targeted drugs, including antibody-based drugs. Moreover, precision medicine has enabled precise diagnosis-based medications, e.g., if the mCRC was EGFR-positive and RAS-wildtype, anti-EGFR antibody drugs such as cetuximab would be effective for cancer. In the present study, we showed that the liquid-based 3D tumoroid assay is markedly useful to evaluate chemotherapeutics' drug efficacy, while presumably of antibody-based drugs as well. To study efficacies of antibody-based drugs, it is recommended to add effector immune cells such as killer T cells, natural killer cells, or macrophages that express Fc receptors and can exert antibody-dependent cellular cytotoxicity (ADCC) or antibody-dependent cellular phagocytosis (ADCP) [25,37,38]. Nevertheless, it is conceived that large tumoroids might be resistant to ADCC and ADCP. We are currently exploring tumoroid-based tumor immunology.

Conclusions
In conclusion, platinum resistance and imatinib resistance in metastatic colorectal cancer were modeled using the liquid-based 3D cultured tumoroid system. The 3D tumoroid system is useful and easily accessible for drug assessment, including chemosensitivity and chemoresistance.
Supplementary Materials: Figure S1: Side population (SP) cells were reduced by verapamil treatment in Colon26 cell line, Figure S2: Interpretation of data.