Rapid generation of high-throughput 3D MTSs using an optimized liquid overlay method
To efficiently and economically prepare size-controllable multicellular tumor spheroids, key factors such as cell seeding density, serum concentrations, external forces, and medium additives were systematically optimized (Table S2). Based on the liquid overlay, cells were seeded on the low-attachment plate coated with non-viscous polymer such that the interaction between cells was greater than the interaction between cells and matrix, and the cells spontaneously gathered into clusters (Costa, de Melo-Diogo, Moreira, Carvalho, & Correia, 2018). The well plates had different shapes, and the effect of tumor cells aggregation in the "U shaped" plates was better than that of the "flat bottom" plates (Table S2). In the "flat-bottom" plates, the cells in each well gathered into multiple cell clusters with uneven shapes and sizes, while in the "U shaped" plates, the cells in each well could aggregate into a single cell spheroid each well under the gravity and interaction of the cells.
Due to the oxygen and metabolite concentration gradients caused by mass transfer limitations, MTSs with a diameter over 500 µm could establish three main cell layers, namely the proliferating cell layer on the surface, the resting cell layer in the middle, and the necrotic core have hypoxic necrosis, which has been reported to as important biological characteristics of MTSs (Hirschhaeuser et al., 2010). The size of MTSs was closely related to the seeding density and the culture time. By recording and observing MTSs every day, we found that the growth of MTSs reached the plateau on the 6th day, on which the MTSs were chosen for the follow-up experiments. The initial seeding density directly determined the final diameter of the MTSs. The diameters of the MTSs on the 6th day were 182.54 µm, 367.21 µm, 512.34 µm, 883.43 µm, 1050.31 µm, with the margin of error no more than 10 %, when the seeding densities were 500, 1000, 2000, 10000 and 20000 cells/well, respectively (Table S2).
Serum contains lots of cytokines, collagen, and adhesion factors, which have significant effects on the growth and proliferation of cells (Brown, Bahsoun, Morris, & Akam, 2019). As shown in Table S2, the low concentration of FBS was detrimental to the formation of MTSs. Under the condition of 5% FBS, the contact between cells was impaired and thus unable to gather into clusters. Compared with 10% serum, increasing the FBS concentration (15%, 20% FBS) could appropriately improve cell aggregation, but was not significant. Considering the cost, 10% FBS was chosen.
The traditional liquid overlay method would promote cells aggregation by introducing external forces to enhance the contact between cells(Costa et al., 2018). Therefore, we investigated the formation and growth of MTSs under the conditions as described in Materials and methods. Compared with the control group, the hanging drop method had poor reproducibility with difficulty to form a single uniform cell spheroid. Besides this, the medium was easy to volatile; EP tube centrifugation significantly improved cell aggregation and reproducibility, which suits lab-scale preparation of 3D MTS for preclinical research, but it was time-consuming and difficult to translate to large-scale production; as for rotary shaker method, the cell clusters were gradually compacted and denser, but it was difficult for the cells to grow and pelletize; by contrast, the whole plate centrifugation significantly improved the roundness and repeatability of MTSs, and it was easy to operate.
To further promote the growth of MTSs, we tested the additives in the culture system. Basement membrane extracts such as Geltrex™ and Matrigel™ contain abundant ECM protein, such as cytokines, laminin and collagen, etc., could promote tumor cell growth, proliferation and invasion, and are usually used in the construction of 3D models for animal cells (Benton, Arnaoutova, George, Kleinman, & Koblinski, 2014). Therefore, we evaluated the effect of 2.5% Geltrex™ and 2.5% Matrigel™ in the formation of MTSs. We found that 2.5% Geltrex™ exerted no significant effect on cell aggregation and growth, while the 2.5% Matrigel™ had extremely obvious promoting effects on the roundness and uniformity of MTSs, and also improved the growth of MTSs (Table S2).
In summary, through the screening of MTSs culture conditions, we have established a rapid MTS formation protocol that allows to produce uniform and controllable MTSs. In brief, the cell suspension was diluted to 104 cells/mL, which was fully mixed with 2.5% (v/v) Matrigel™, and then a volume of 200 µL of this cell suspension was added to each well of "U-shaped" low-attachment 96-well plates. The spheroid formation was initiated by centrifuging the plate at 1000 g for 10 min. Afterwards, the plates were cultured in an incubator at 37°C with 5% CO2 and 95% humidity for 6 days. In this study, the protocol was exactly followed in the formation of multicellular tumor spheroids, unless otherwise specified. In addition to Hela cells used in the follow-up study, other tumor cell lines (colon cancer cells HCT116, liver cancer cells HepG2, breast cancer cells MCF-7, lung cancer cells A549, bladder cancer cells 5637) and normal cell line (human hepatocytes L-02) can form reproducible cell spheroids following this protocol (Figure. S1). Not only this, this protocol can be extended to co-culturing model with stromal cells (such as fibroblasts UCF, immune cells PBMC) to construct multi-component MTSs (Figure. S1).
Physiological characterization of 3D MTSs
To monitor the growth dynamics of MTSs, the diameter, roundness and cell growth of MTSs were recorded and assessed every day. In the initial stage of MTSs formation, cells grew slowly and were easy to distinguish individual cells (Figure. 2A). Driven by the interaction and contact between the cells, cells gradually gathered into clusters. After 2 days of culture, the cells entered the logarithmic growth phase, proliferated rapidly and gradually formed smooth surfaces. On the 6th day, MTSs reached the plateau stage, with the largest diameter of 532.21 ± 21.42 µm and 592.54 ± 13.64 µm, respectively, when the seeding densities were 1000 and 2000 cells/well (Figure. 2B). The roundness of obtained MTSs exceeded 0.9 (Figure. 2C), with the margin of error no more than 10%. The number of living cells in a single MTS reached the maximum on day 6, approx. 17 times the initial amount (Figure. 2D). After that, the boundaries of MTSs began to wrinkled and blurred, and the spheroids gradually disintegrated.
The morphology and microstructure of MTSs cultured for 6 days were observed by the SEM, and the MTSs showed a good 3D structure and regular spherical shape (Figure. 2E, Figure. S2). Interestingly, MTSs of different cell types had different surface structures. The surface of the single-component Hela cell spheroid was relatively smooth (Figure. 2E), while the multi-component MTSs retained the more obvious tissue structures (Figure. S2). When co-cultured with fibroblasts such as UCF, the fibroblasts extended outwards MTSs. Obvious vesicles were observed in the MTSs co-cultured with immune cells such as PBMC. This indicates that the more complex 3D model would preserve more completely biological characteristics of tumor tissues in vivo.
Compared with 2D monolayer cells, the cell proliferation rate in 3D MTSs was reduced. The specific growth rate (µ) of cells cultured for 2 days under 2D conditions was 0.0455 h− 1, while the µ of cells in MTSs cultured for 2, 4, and 6 days were 0.0263 h− 1, 0.0246 h− 1, 0.00522 h− 1, respectively, with the margin of error no more than 13.6 % (Figure. S3). The cellular viability of 3D MTSs that were cultured for 2 days was similar to 2D monolayer cells, but the percentage of living cells and apoptotic/ necrotic cells in MTSs are gradually decreased and increased, respectively, with the culture time (Figure. 2F). Compared with 2D monolayer culture, the cell cycle in 3D MTSs was blocked throughout the cultivation. In addition, cells in the 3D MTS are accumulated and decreased over the culture age in the G1 and S phase, respectively (Figure. 2F). However, there was no significant change in the proportion of cells in G2 phase between 2D monolayer culture and 3D MTSs.
Chemosensitivity of Hela tumor spheroids following 5-FU treatment
The cytotoxicity of 5-FU to Hela cells cultured under both 2D monolayer and 3D MTSs was evaluated (Figure. 3A). 3D MTSs showed stronger resistance than 2D monolayer cells, and after the 5-FU treatment for 48 h the IC50 of MTSs (93.88 µM, 95 % confidence interval: 64.88 ~ 148.5 µM) was approximately 5.72 times that of 2D monolayer culture (16.42 µM, 95 % confidence interval: 12.73 ~ 21.14 µM). Therefore, we treated the HeLa carcinoma cells from both 2D and 3D models with 16 µM 5-FU for 48 h in the following study. Next, the changes of cell apoptosis and cell cycle were compared before and after the 5-FU treatment (Figure. 3B). The effect of 5-FU on cells in 3D MTSs was significantly reduced compared with 2D monolayer culture. After the 5-FU treatment, the percentage of living cells and apoptotic cells was decreased and increased, respectively, in 2D monolayer, while there were no significant changes in 3D MTSs. As is known, 5-FU mainly targeted S-phase cells (Ijichi, Adachi, Ogawa, Hasegawa, & Murakami, 2014). After the 5-FU treatment, the cell cycle was blocked in G1 phase in both 2D monolayer and 3D MTSs. Meanwhile, we observed a decrease in the proportion of S phase cells and a constant proportion of G2 phase cells in the 2D monolayer after the 5-FU treatment, while no significant changes were observed in the 3D MTS.
To adapt to rapid metabolic requirements of tumor cells, the metabolic pathways have significantly changed to metabolize nutrients in a manner conducive to proliferation rather than efficient ATP production (Seyfried, Arismendi-Morillo, Mukherjee, & Chinopoulos, 2020). Mitochondrion is the energy factory and the main position of oxygen consumption, and the abnormal mitochondrial metabolism in tumor cells are often related to drug resistance (Yan & Li, 2018). According to the mitochondrial respiration profile (Figure. 3C), the oxygen consumption in 3D MTSs was much higher than that of 2D monolayer cells, which was mainly due to the significant increase of non-mitochondrial respiration. The results showed that the non-mitochondrial respiration of MTSs was about 4.60 and 3.19 times that of 2D cultures under the control and 5-FU treatment conditions, respectively. The possible reason for the increased non-mitochondrial respiration was that the main energy source shifted from mitochondrial oxidative phosphorylation to glycolysis. Contrary to this, the spare respiratory capacity of MTSs was reduced about 74.98 % and 63.31 relative to the 2D culture, which was associated with mitochondrial dysfunction. Furthermore, the mitochondrial basal respiration and maximum respiration capacity were reduced by 68.56 % and 71.81 % in 3D MTSs relative to the 2D culture under the control conditions, while the ATP synthesis capacity was almost constant. After the 5-FU treatment, the mitochondrial basal respiration and maximum respiration capacity was almost constant, while the ATP synthesis capacity were about 62.40 % reduced in 3D MTSs as compared with the 2D cultures. The further inhibition of 5-FU on the mitochondrial ATP production capacity of 3D MTS aggravated the dependence of MTS on the glycolytic pathway. As the major hub of cellular energy generation, mitochondrion, is also the main source of reactive oxygen species (ROS) and it has been reported that the ROS level can be reduced if glycolysis as the main energy source (Herst, Tan, Scarlett, & Berridge, 2004). However, we found that the ROS production capacity in 3D MTSs was significantly higher than 2D monolayer cells (Figure. 3D).
As can be seen from Figure. 3E, 2F, in the presence of glucose and glutamine, tumor cells preferred to use glutamine. Under control conditions, cells cultured in 2D monolayer cultures and 3D MTSs mainly used glutamine, and began to consume a small amount of glucose after 60 h of culture. Under the 5-FU treatment, glutamine was still the main energy source in 2D monolayer culture, while cells consumed glutamine and glucose at the same time, in 3D MTSs. In addition, with 5-FU treatment reabsorption phenomenon was found in the 3D MTSs, while HeLa cells continued to secrete ammonia under 2D monolayer culture conditions (Figure. 3G). It has been reported that the reabsorption and reuse of ammonia was beneficial to the growth of tumor cells (Spinelli et al., 2017). The secretion rate of lactate in 3D MTSs was slightly higher than that in 2D culture under both control and 5-FU treatment conditions (Figure. 3H, Figure. S4).
Identification of transcriptional alterations for 5-FU resistance
The biological characteristics of 3D MTSs were more representative than that of 2D monolayer culture, such as hypoxic regions and pH gradients caused by mass transfer limitations, enhanced ECM secretion, drug permeation barriers caused by closely cells contact, increased ECM deposition and the improvement of tumor cell stemness (Dittmer & Leyh, 2015). Therefore, we measured the transcriptional levels of drug resistance-related genes in Hela cells cultured under both 2D cell cultures and 3D MTSs (Figure. 4).
In the 3D MTS model, there is often an oxygen diffusion limit of 150 ~ 200 µm (Oldham, Clish, Yang, & Loscalzo, 2015). Hence, exceeding this radius would likely form an area of hypoxia in the MTSs, which would lead to genetic and metabolic reprogramming regulated by hypoxic induction factor (HIF1A), enabling tumor cells to acquire a more aggressive phenotype (Denko & Nicholas, 2008; Zhang et al., 2020). Compared with 2D monolayer cells, after the 5-FU treatment, the transcript level of HIF1A in the 3D MTSs were significantly increased by 1.60 times (Figure. 4). It has been reported that tumor hypoxia microenvironment would cause abnormal activation of the oncogene MET (Stella, Benvenuti, & Comoglio, 2010), and consistent with this, the transcription level of MET was 2.54 and 1.8 times higher up-regulated in the 3D MTSs than in the 2D cultures under the control and 5-FU treatment conditions, respectively, which would promote angiogenesis and maintain tumor aggressiveness. HIF1A could also induce the expression of vascular endothelial growth factor (VEGFA), which is central to the growth and metastasis of tumors, thereby promoting the malignant progression of tumors (Mohamed, Khalil, & Toni, 2020). As evidenced, we observed that the transcription level of VEGFA was about 5.04 times and 14.87 times higher up-regulated in more 5-FU resistant 3D MTSs than 2D monolayer cells under the control condition and the 5-FU treatment conditions, respectively.
Compared with normal cells, tumor cells display upregulated glycolysis for the provision of intermediates for rapid proliferation, which is mainly manifested by enhanced glucose uptake and lactate excretion (Yong, Stewart, & Frezza, 2019). Similarly, we observed an increase in glucose utilization and lactate excretion in 3D MTSs, especially following the 5-FU treatment, compared with 2D monolayer culture (Figure. 3E, 2H). Therefore, the transcript level of the genes encoding glucose transporter (GLUT), lactate dehydrogenase (LDH) and phosphofructokinase (PFK1) were measured. As compared with the 2D monolayer cultures, the transcript levels of GLUT1, LDHA, PFK1 were 4.20, 1.60 and 1.21 times higher upregulated in the 3D MTSs under the control conditions (Figure. 4). As expected, the transcript levels of these genes were more pronouncedly increased after the 5-FU treatment, showing that 9.72, 2.45 and 1.88 times higher upregulated in the 3D MTSs than in the 2D cultures. This result indicated that 3D MTSs featured enhanced aerobic glycolysis, i.e., the well-known Warburg effect.
The rapid proliferation of tumor cells required a large number of nucleic acids. The only source of thymine in cells is the de novo synthesis pathway, and the high expression of thymidylate synthetase (TYMS) is often associated with poor prognosis of tumors (Donner et al., 2019). 5-FU blocks DNA synthesis to induce cell death by inhibiting TYMS, while dihydropyrimidine dehydrogenase (DPYD) could decompose and deactivate 5-FU before it was converted into active metabolites (Negarandeh et al., 2020). Compared with the 2D cultures, the transcription level of TYMS were 2.15 and 3.48 times higher up-regulated, meanwhile the transcript level of DPYD were 3.66 and 1.26 times higher up-regulated in 3D MTSs before and after the 5-FU treatment conditions, respectively (Figure. 4). Therefore, the up-regulation of the expression of TYMS and DPYD were also the reasons for the enhanced resistance of MTSs to 5-FU.
Increasing evidence has ever shown that conventional cancer chemotherapy is seriously limited by the multidrug resistance (MDR) commonly exhibited by tumor cells (Perez-Tomas, 2006). In drug-resistant tumor cells, the main mechanism was the drug accumulation and efflux in which the ATP binding cassette (ABC) transporters played an important role (Orlando & Liao, 2020; Ye et al., 2016). However, we did not observe the up-regulation of ABCB1 and ABCG2 transcription levels in 3D MTSs as expected, but there may be differences in protein or metabolite levels. Lysosome-associated transmembrane protein 4B (LAPTM4B), a multidrug resistance gene, could stimulate drug resistance and promote cell growth and proliferation by regulating drug efflux mechanism and activating PI3K/Akt signal transduction (Gu et al., 2020). Consistent with this, compared with the 2D monolayer cultures, the expression level of LAPTM4B was 1.57 and 1.32 times higher up-regulated in 3D MTSs before and after the 5-FU treatment (Figure. 4).
Apoptosis defects were also one of the reasons for drug resistance, which were usually regulated by the BCL-2 protein family (Warren, Wong-Brown, & Bowden, 2019). Compared with 2D monolayer cells, the transcription level of the BCL2 encoding anti-apoptotic protein in 3D MTSs was 1.55 times and 3.09 times higher up-regulated before and after the 5-FU treatment, respectively, which contributed to the progression of tumor drug resistance (Figure. 4). However, the transcription level of the BAX encoding apoptotic protein was also up-regulated in 3D MTSs, which might be related to the decrease of cell proliferation activity.
Then, the expression of cytokines related to tumor cell proliferation and progression was tested. Transforming growth factor (TGFB1) was generally up-regulated in tumor cells, and could induce epithelial-mesenchymal transition and promote tumor cell growth, proliferation and invasion (Fuxe & Karlsson, 2012). The mTOR pathway was a classical signal transduction pathway regulating cell growth and metabolism, and was dysregulated in many cancers (Caron et al., 2010). It has been reported that the glycolytic pathway was affected by the mTOR pathway through two key transcription factors, HIF1A and MYC (Renner et al., 2017). As compared with the 2D monolayer cultures, the transcript levels of TGFB1, MTOR, MYC were 1.92, 2.08 and 1.59 times higher upregulated in the 3D MTSs, respectively (Figure. 4). As expected, the transcript levels of these genes were more pronouncedly increased after the 5-FU treatment, showing that 1.82, 3.21 and 3.51 times higher upregulated in 3D MTSs than in the 2D monolayer cultures, respectively (Figure. 4).
ECM, such as laminin, fibronectin, vimentin, mediated interactions between cells and participated in signal transduction in processes such as cell adhesion, migration, invasion, proliferation and EMT to promote the development of drug resistance, referred to as cell adhesion mediated drug resistance (CAM-DR) (Baltes et al., 2020; Valkenburg, de Groot, & Pienta, 2018). Although the transcript level of laminin β1 (LAMB1) did not change significantly, compared with the 2D cultures, we found 3.68 times higher up-regulation in the expression of fibronectin (FN1) in 3D MTSs under the control condition. Meanwhile, the transcript levels of vimentin (VIM) were 1.70 and 3.18 times higher upregulated in 3D MTSs under the control and 5-FU treatment conditions, respectively (Figure. 4). It was found that ECM participated in CAM-DR by stimulating integrin mediated PI3K activation to protect tumor cells from damages caused by radiotherapy and chemotherapy(Mohanty et al., 2020) (Hodkinson, Mackinnon, & Sethi, 2007). Compared with the 2D monolayer cultures, the expression level of integrin β1 (ITGB1) was 1.76 and 3.59 times higher up-regulated in 3D MTSs before and after the 5-FU treatment, respectively (Figure. 4). ITGB1 (also known as CD29) and CD44 are reported as tumor stem cell markers, and the presence of tumor stem cells could affect the drug treatment and subsequent tumor recurrence (Tomasetti, Li, & Vogelstein, 2017). It has been shown that CD44 can mediate the stemness of tumor cells and participate in metastasis by binding to hyaluronic acid (Gomez et al., 2020). Compared with 2D monolayer cultures, the expression level of and CD44 was 1.79 and 2.21 times higher upregulated in 3D MTSs than in 2D monolayer culture under the control and 5-FU treatment conditions, respectively (Figure. 4).
Proteome analysis
Tumor cells synthesize a variety of proteins that interact with each other to perform cellular functions through transcriptional, translational and post-translational modifications (Manzoni et al., 2018). Genetic variation and changes of the tumor microenvironment would affect the expression, structure and interaction of proteins, leading to changes in the activities of cells. The differences in protein expression levels between 2D monolayer cultures and 3D MTSs cultures before and after the 5-FU treatment were determined through Tandem Mass Tag (TMT) technology. The principal component analysis (PCA) result showed that there were significant differences among all sample groups and good repeatability within the sample group (Figure. S5A), which proved that the obtained experimental data could support the following data analysis. A total of 5262 proteins were determined, and the number of differential proteins was shown in Table 1. We found that the effect of 16 µM 5-FU on the expression profile of protein in 3D MTSs was less pronounced than that in 3D monolayer cells. Under the 5-FU treatment, there were 133 proteins significantly different in 2D monolayer cells compared with the control condition, while only 27 differential proteins were significantly changed in the 3D MTSs, which indicated that 5-FU induced response in 3D MTS was attenuated. Our results showed that, between the control condition and 5-FU treatment, there were significant differences in cellular processes related to cell growth and death (e.g., cell cycle, apoptosis and senescence) under 2D monolayer culture (Figure. S5B). However, the differential proteins in the 3D MTSs were not enriched in specific metabolic pathways. While upregulating the negative cell cycle regulatory protein RB1 to block the cell cycle, 2D monolayer cells upregulated cyclin (CCNB1, CCNB2), apoptosis inhibitor BIRC5 to maintain growth activity with the 5-FU treatment (Figure. S5C). In the 3D MTSs, we also observed the up-regulation of BIRC5 with the 5-FU treatment. Meanwhile, 5-FU treatment reduced the absorption and transport of glucose, folate and amino acids in 2D monolayer cultures, and down-regulated the members of solute carrier family members, including SLC2A1, SLC19A1, SLC38A1 and SLC38A2, which were not significant in the 3D MTSs (Fig S5C).
Table 1
The number of differential proteins between 2D monolayer culture and 3D MTS before and after the 5-FU treatment.
Control group | Experimental group | Up | Down | Sum |
2D Monolayer | 3D MTS | 206 | 161 | 367 |
2D Monolayer + 5-FU | 3D MTS + 5-FU | 185 | 103 | 288 |
2D Monolayer | 2D Monolayer + 5-FU | 51 | 82 | 133 |
3D MTS | 3D MTS + 5-FU | 8 | 19 | 27 |
Table 1 shows 367 and 288 differential proteins between 2D monolayer cultures and 3D MTS before and after the 5-FU treatment, respectively. Based on this, we selected 50 differential proteins which exerted great influence on the growth, metabolism and drug resistance of tumor cells, and then their expression levels were hierarchically clustered between 2D monolayer cultures and 3D MTSs before and after the 5-FU treatment. The differential proteins between 2D monolayer cultures and 3D MTSs displayed similar profiles (Figure. 5A). In order to further analyze the metabolic pathway differences, KEGG pathway enrichment analysis was performed with all differential proteins between 2D monolayer cultures and 3D MTSs (Figure. 5B). Processes related to protein synthesis, processing and transportation, mainly in heat shock protein family, exhibited the most significant differences between 2D monolayer cultures and 3D MTSs. In addition, there were significant differences in the mutual conversion of amino acids, in which mitochondrial glutamate oxaloacetic acid transaminase (GOT2) was significantly down-regulated in 3D MTSs. GOT2 plays an important role in amino acid metabolism and TCA cycle (Yang et al., 2015). Consistent with the transcription level (Figure. 4), we found more multidrug resistance-related proteins (ABCC4) and ECM proteins in 3D MTSs, including laminin (e.g., LAMA5) and collagen (e.g., COLO1A1). In the control condition, LAMA5 and COLO1A1 was upregulated in 3D MTSs compared with 2D monolayer cultures (Figure. 5A). After the 5-FU treatment, 3D MTSs upregulated more kinds of ECM protein, including LAMA1, LAMB1, LAMB2, LAMC1 and COL4A1. Similarly, under the 5-FU treatment, we also found that integrin α5 (ITGA5) was up-regulated in 3D MTSs (Figure. 5A), which was consistent with the up-regulated transcription level of ITGB1 (Figure. 4). Notably, glucose transport related proteins, such as SLC2A1, were significantly up-regulated, which further proved the possibility of up-regulation of glycolytic flux of 3D MTSs. In KEGG pathway enrichment analysis, we found that signaling pathways such as PI3K/Akt and HIF-1 were significantly altered after the 5-FU treatment (Figure. 5B). In this study, HYOU1 in 3D MTSs was also significantly up-regulated regardless of the 5-FU treatment (Figure. 5A). Hypoxia up-regulated 1 (HYOU1), a member heat shock protein 70 family, maintains endoplasmic reticulum (ER) homeostasis under hypoxia conditions while promotes the growth, metastasis and invasion of tumor cells by activating the PI3K/Akt signaling pathway (Li et al., 2019).
Through GO analysis, the gene functions of the differential proteins were analyzed (Figure. 5C, Table S3 and Table S4). Under both control and 5-FU treatment conditions, the differential proteins between 2D monolayer cultures and 3D MTSs were similar in enrichment levels. The molecular functions of 3D MTSs were mainly performed by up-regulating proteins related to calcium ion binding, ECM constituent, integrin binding, and down-regulating proteins related to ATP binding, RNA binding, ubiquitin protein ligase binding. While for the biological process, 3D MTSs mainly up-regulated processes related to ECM organization, cellular protein metabolic process tissue development, blood vessel development and protein folding in endoplasmic reticulum, and down-regulated processes related to mitochondrial function, including mitochondrial translation, mitochondrial transcription and mitochondrial electron transport, etc. In addition, with the 5-FU treatment, MTSs also up-regulated proteins related to cell adhesion and down-regulated proteins related to cell division and cell redox homeostasis. On the other hand, for cellular components, the proteins up-regulated in MTSs were mainly distributed in ER or extracellular regions such as exosomes, ECM, basement membrane, while the down-regulated proteins were mainly distributed in the mitochondria. This is reasonable that the increase of the production and deposition of ECM enhanced the physical barrier of drug penetration, as well as involved in the regulation of the EMT process and the expression of tumor cell stem genes, thereby promoting the development of tumor resistance (Dominijanni, Devarasetty, & Soker, 2020; Joyce et al., 2018). Mitochondrion, an important signal transduction hub, can regulate cell apoptosis and participate in cell communication and tumor formation through ROS, nitric oxide, and calcium ions and proteins involved in the apoptotic cascade (Frezza, 2014). It has been reported that the down-regulation of mitochondria-related genes promoted the up-regulation of EMT pathway and promoted the invasion and metastasis of tumor cells (Gaude, 2018). Therefore, the up-regulation of ECM construction and the down-regulation of mitochondrial function genes might lead to the enhanced resistance of 3D MTSs to 5-FU. The interactions between differential proteins were further analyzed using the STRING database for the protein-protein interaction network (Figure. 5D), we found that the molecular chaperones related to protein folding, assembly, and secretion exerted the most obvious effects, and the heat shock protein family was dominant (such as, HSPA5, HSP90B1, HSPG2). The increased expression of molecular chaperones was conducive to maintaining the growth of tumor cells in unfavorable environments, and the overexpression of HSP90 is related to the poor prognosis response of tumors and the increase of drug resistance (Jarosz, 2016).
Metabolome analysis
The metabolome could directly reflect the physiological characteristics of cells, and the minor changes in gene transcription and protein expression would cause significant differences in the metabolome (Klein & Heinzle, 2012). Tumor cells often alter signal transduction pathways and rewire metabolism, thereby promoting cell proliferation and weakening the effect of anti-cancer drugs (Locasale, 2012). In this study, we analyzed 330 metabolites, including carbohydrates, amino acids, lipids, nucleotides, organic phosphates, sterols, coenzymes of tumor cells cultured under both 2D monolayer cultures and 3D MTSs before and after the 16 µM 5-FU treatment. The PCA result showed the significant differences among groups and good repeatability in groups, which proved that the following data analysis could be supported (Figure. S6A). Table 2 shows 107 and 76 differential metabolites under 2D monolayer cultures and 3D MTS before and after the 5-FU treatment, respectively. With the 5-FU treatment, the metabolites involved in the ABC transport system of cells were significantly different between 2D monolayer cells and 3D MTSs (Figure. S6B, Figure. S6C). The effects of 5-FU on sugar metabolism and amino acid metabolism in 3D MTSs were less pronounced than in monolayer cell culture. Under the 5-FU treatment, the intermediates of TCA cycle (including citrate, malate and fumarate) and of alanine, aspartate and glutamate metabolism were all decreased in 2D monolayer cells, while there were no significant differences in 3D MTSs, except for the increase of asparagine. Similarly, the effect of 5-FU on the nucleotide metabolism of MTSs was also relatively weakened. 5-FU disrupts the de novo nucleotide synthesis pathway by targeting TYMS (Longley et al., 2003). Under the 5-FU treatment, the nucleotide synthesis pathway of 2D monolayer cells was inhibited with the increase of 5-phospho-ribulose, while purine nucleotide synthesis was blocked with the increase of purine synthesis precursors inosine and hypoxanthine and the decrease of inosine monophosphate (IMP) and adenosine monophosphate (AMP). It was also found that the precursor orotate to uracil synthesis was decreased in 2D monolayer cells with the 5-FU treatment. Under the 5-FU treatment, we find the decrease of inosine, xanthine and orotate, and the increase of uridine. After 5-FU treatment, the pentose phosphate pathway (PPP) changed significantly. Under 2D culture conditions, gluconolactone and gluconic acid were significantly increased, while 6-phosphogluconic acid was significantly decreased, possibly due to the increased production of ribulose 5-phosphate (Figure. S6B). This process is accompanied by the production of NADPH (Cairns, Harris, & Mak, 2011), which contributes to the elimination of ROS induced by 5-FU (Figure. 3D). In 3D MTSs, we also observed an increase in ribulose 5-phosphate, which resulted in a significant decrease in gluconolactone and gluconic acid (Figure. S6B).
Table 2
The number of differential metabolites between 2D monolayer culture and 3D MTS before and after the 5-FU treatment.
Control group | Experimental group | Up | Down | Sum |
2D Monolayer | 3D MTS | 44 | 55 | 99 |
2D Monolayer + 5-FU | 3D MTS + 5-FU | 56 | 35 | 91 |
2D Monolayer | 2D Monolayer + 5-FU | 27 | 80 | 107 |
3D MTS | 3D MTS + 5-FU | 41 | 35 | 76 |
In the control conditions, there were 99 different metabolites between 2D and MTSs while 91 different metabolites were detected under the condition of the 5-FU treatment. We selected 50 differential metabolites which exerted greater impacts on tumor cell growth, metabolism, and drug resistance, and then their expression levels were hierarchically clustered between 2D monolayer cultures and 3D MTSs under both control and 5-FU conditions (Figure. 6A). KEGG enrichment was performed to examine the differences in metabolic pathways of tumor cells under different culture conditions (Figure. 6B). The ABC transport system played an important role in metabolite transport and was associated with the multidrug resistance (Nunes, Costa, Barros, de Melo-Diogo, & Correia, 2019). We found that the metabolites involved in the ABC transport system, were significantly different between 2D monolayer cultures and 3D MTSs regardless of the 5-FU treatment, which may indicate the differences in drug transport capacity. The carbon metabolism pathways performed significant differences in 2D monolayer cultures and 3D MTSs. Under control conditions, compared with monolayer cells, the TCA cycle intermediates including citric acid, cis-aconitic acid, succinic acid, fumaric acid, malic acid were decreased in 3D MTSs, which was consistent with the significant increase in cellular non-mitochondrial respiration and the down-regulated protein expression related to mitochondrial activity (Figure. 3C, Table S3 and Table S4). However, we did not observe significant differences in the intermediate metabolites of the glycolysis pathway between 2D monolayer cultures and 3D MTSs. This indicated that the flux control may be governed by other regulation mechanisms, e.g., phosphorylation-mediated reprogramming of glycolytic activity (Ruprecht et al., 2017). We found that there were no significant differences in glutamine and glutamate between 2D monolayer cultures and 3D MTSs, while with 5-FU treatment, glutamine and glutamate significantly increased in 3D MTSs (Figure. 6A). Glutamine participated in the TCA cycle by converting to glutamate and then further to α-ketoglutarate, which maintained the TCA cycle in 3D MTSs under the 5-FU treatment, resulting in no significant difference in the TCA cycle between 2D culture and 3D MTSs after 5-FU treatment. Meanwhile, asparagine significantly decreased under both conditions in the 3D MTSs, which led to the decrease of aspartate involved in the TCA cycle. Furthermore, compared with 2D culture, gluconolactone, gluconic acid and ribulose 5-phosphate involved in the PPP were significantly increased in 3D MTSs under the control condition; under the 5-FU treatment, only the significant increase of ribulose 5-phosphate was found in 3D MTSs.