Organoid technology and applications in cancer research

During the past decades, enormous efforts have been exerted to cancer research [1, 2] and substantial progresses have been achieved in diagnosis [3, 4] and treatment [5–12]. However, cancer still represents a major worldwide health concern and many obstacles remain to be solved for further improving life quality and prolonging survival of cancer patients. The development of effective treatment regimens is among the major hurdles. Due to poor recapitulation of human tumors by conventional cancer models, numerous drugs working in these cancer models are finally eliminated in clinical trials because of either ineffectiveness or unbearable side effects.

Traditional two-dimensional (2D) cell line cultures and patient-derived tumor xenografts (PDTXs) have long been employed as tumor models and have made tremendous contribution to cancer research. However, many drawbacks hamper these two models for clinical use. 2D cell line cultures show their inability in simulating some vital subjects, such as the immune system, microenvironment, stromal compartments, and organ-specific functions. Other limitations include the lack of genetic heterogeneity of original tumors after many passages for cancer cell lines [13] as well as experiencing mouse-specific tumor evolution [14] and being consuming in money, time, and resources for PDTXs [15].

Organoid technology springs up and becomes an independent research tool. Organoids are three-dimensional (3D) constructs and can be developed from embryonic stem cells (ESCs), induced pluripotent stem cells (iPSCs), somatic SCs, and cancer cells in specific 3D culture system (Fig. 1). Stem cells are a class of under-differentiated cells with self-renewing capacity and the potential to regenerate various tissues and organs. According to the developmental stage in which stem cells are located, they are divided into embryonic stem cells and adult stem cells. Embryonic stem cells are a type of cells isolated from early embryos with the ability of unlimited proliferation, self-renewal, and multi-directional differentiation. Progenitor cells belong to adult stem cells and are undifferentiated pluripotent or multipotent stem cells. Progenitor cells are present in various adult tissues of organisms and are responsible for the repair and regeneration process after tissue damage.

These amazing 3D tissues in small scale are fabricated in the laboratory and resemble the parent organ in vivo in terms of structure and function. Three basic features are as follows: firstly, it contains multiple cell types of the in vivo counterpart; secondly, the cells organize similarly to the primary tissue; thirdly, it functions specifically to the parent organ [16]. This powerful technology bridges the conventional 2D in vitro models and in vivo models, and exerts great potential for clinical applications (Fig. 2), especially in cancer research [17]. Tumor modeling might be a pivotal branch of organoid technology [18, 19], including modeling infection-cancer development [20, 21], mutation-tumorigenesis processes [22, 23] and genetic carcinoma [24, 25]. Apart from cancer modeling, organoid technology also exerts enormous potential in evaluation of efficacy and toxicity of drugs [26], regeneration medicine [27, 28], and precision treatment [29, 30]. Until quite recently, organoids have been established successfully for multiple cancer types, including stomach cancer [26], colorectal cancer [31–33], liver cancer [34], pancreatic cancer [35, 36], prostate cancer [37], and breast cancer [38].

In this review, we outline a brief history of organoids, describe organoids of diverse cancer types, focus on the potential applications of this promising technology in oncology, and finally discuss the current limitations.

The notion that mammalian cells are inherently endowed with self-organizing capacity has long been widely known among researchers, and this ability has been employed to develop 3D cultures from primary tissues. Numerous types of culture systems have been reported in early studies [39–42], but no methods could achieve long-term culture and maintain the basic crypt-villus physiology. Encouragingly, the year 2009 witnessed the advancement of intestinal organoid culture system, a chief technological breakthrough in the SC field [43]. The novel culture system contained laminin-rich Matrigel replacing extracellular matrix (ECM) and growth factors including epidermal growth factor (EGF), Noggin, Wnt, and R-spondin. 3D mouse crypt structures in which continuously renewing epithelial layer exhausted apoptotic cells into a central lumen lined by crypt-like and villus-like sections were established in this 3D culture system, and these features remained when cultured for 8 months [43]. Subsequently, this culture system was adapted for the establishment of human intestinal organoids and other organ 3D architectures, such as the liver, stomach, and colon [44–46]. The organoid technology has become widely accepted in recent years, since these 3D cultures faithfully recapitulate the genotype, phenotype, and cellular behaviors of parent tissues [47].

Breast organoid cultures also experience a gradual evolution from the earliest attempts of in vitro cultivation of organ explants to the current relatively refined versions [48–50]. Mammary gland explants of from virgin mice could be cultivated in a serum-free medium, which consists of four major components: aldosterone, prolactin, insulin, and cortisol [48]. Through testing the mammary-derived growth inhibitor (MDGI) in mammary explants in vitro from mice at different development stages, it was demonstrated that MDGI expression was correlated with functional differentiation of normal mammary gland [48]. Next, mouse mammary gland cultivated in organ culture containing MRG protein showed a differentiated morphology with the upregulation of beta-casein [49]. Recently, it has been indicated that 3D cultures of breast cancer could more accurately model the structural and functional changes during the conversion from breast ductal carcinoma in situ to invasive carcinoma [50]. Up to now, breast cancer organoids have been efficiently established for studying breast cancer biology, and efforts are still in need for further improving culture conditions in order to overcome the current limitations.

Early in the 1980s, organotypic cultures have been employed to cultivate embryonic kidney, which allowed accurate manipulation of diversity developmental events in vivo in comparison with monolayer cell cultures [51]. However, the in vitro conditions led to metabolic changes, and it was difficult to realize long-term cultures because of the nutrition insufficiency-induced tissue damage [51]. When fetal murine metanephric tissues were isolated and incubated in serum-free medium, organotypic proximal tubular and glomerular epithelial differentiation were observed but without perfusion, urine production, and vascularization [52]. Quite recently, it was reported that host-derived vascularization formed in iPSC-derived kidney organoids in fully defined conditions without any exogenous vascular endothelial growth factor [53]. Progressive morphogenesis, including functional glomerular perfusion in function as well as connection to pre-existing vascular system, glomerular basement membrane, and fenestrated endothelial cells in structure, was observed in these organoids after transplanted under the kidney capsule [53].

Isolated brain cells, cultured in serum-free medium with classical hormones, EGF, fibroblast growth factor (FGF), attachment factors/basal membrane components, transport proteins, transferrin, albumin, vitamins, experienced morphological, bioelectrical, and biochemical differentiation [54–56]. During the past a few years, a variety of neural organoids have been established from ESCs or iPSCs in refined 3D culture systems which faithfully manipulated brain structures and some specific functions, including the whole brain [57] and sub-brain regions, such as hypothalamus [58], adenohypophysis [59], midbrain [60], cerebellum [61], and hippocampus [62].

Some infectious pathogens are identified to be significant risk factors of cancer, such as Helicobacter pylori in gastric cancer, Salmonella enterica in gallbladder carcinoma, hepatitis virus in HCC, and Epstein-Barr virus (EBV) in gastric cancer, nasopharyngeal carcinoma, and lymphoma. However, there is still a lack of extensive understanding of the direct relationships and causal mechanisms between the infectious pathogens and corresponding cancers. Organoids can serve as a potential excellent model for studying these processes through co-culture systems with different pathogens. Neefjes J and colleagues employed co-cultures of murine-derived genetically predisposed gallbladder organoids and Salmonella enterica to explore the epidemiological association between gallbladder carcinoma and Salmonella Typhi infection, and supported that Salmonella enterica triggered and maintained malignant transformation accompanied by TP53 mutations and c-Myc amplification through Salmonella enterica effectors-induced activation of mitogen-activated protein kinase and AKT pathways [20]. Besides, viral infectious organoid models can also be established as exemplified by intestinal organoids with rotavirus infection [21], indicating that the virus-tumor relationship can also be simulated by co-culture systems, such as hepatitis virus versus liver cancer and EBV versus nasopharyngeal carcinoma. Modeling of the transition from infection to tumor formation and progression of organoids might help to reveal pathogenic mechanisms and find potential anti-tumor targets during this process.

Cancers occur on the genetic basis of sequential accumulation of mutations, signifying that it is pivotal to throw light upon the mutational processes during homeostasis and tumorigenesis. Knowledge of original mutation profile has been demonstrated to be of importance [22], for which healthy organoids provide a platform. Whole genome sequencing on human colon organoids with knockout of DNA repair genes through CRISPA-Cas9 technology revealed that the deficiency in mismatch repair genes contributed to mutation accumulation through replication errors, and deficiency in the cancer-predisposition gene DNA glycosylase led to mutation profile previously noted in cancer patients [23]. In addition, understanding of heterogeneous mutational signatures underlying tumor progression is also of great significance, which can also be prompted by organoid technology. Remarkably increased mutation rates and acquisition of new mutational profile were observed during development of colorectal tumoroids, and the diverse contributions of mutational processes in different regions of the same tumor were demonstrated by Roerink SF and colleagues [87]. It is interesting and feasible to employ organoid platform to evaluate the impact of drugs and irradiation on mutation profiles of cancer and normal cells as well as explore the mutational differences between sensitive and resistant organoids towards treatments.

Genetic cancer modeling is another paramount potential application of tumoroids [24, 25, 88, 89]. The conversion from healthy human intestinal organoids to colorectal progressive tumoroids has been achieved through the introduction of a set of common driver mutations in colorectal cancer via CRISPR-Cas9 gene editing technology, indicating tumor growth as a consequence of cancer driver mutations was independent of SC niche factors and identifying loss of adenomatosis polyposis coli (APC) and TP53 as pivotal contributors for chromosome instability and aneuploidy [24, 90]. Using organoid models, it was demonstrated that ring finger protein 43 mutations positively regulated Wnt-β-catenin signaling in human serrated colon adenoma [91], and loss of mutations in caudal type homeobox2 and BRAF^V600E synergistically drove progression of serrated colorectal cancer [89]. Organoids facilitate better understanding of tumor initiation and progression of cancers at the genetic level.

During the past decades, numerous anti-cancer drugs developed from screening on conventional 2D culture of large standard cell lines failed in clinical studies [92, 93]. For most cytotoxic agents, broad activity was observed across tumor cell lines, but clinical efficiency noted in patients was in more limited settings [93]. Voskoglou-Nomikos T evaluated whether in vitro cell lines were reliable in predicting clinical utility. The results showed that in vitro cell line model was predictive for non-small cell lung cancer under the disease-oriented approach, but not for colon cancer [94]. Since cancer organoids are near-physiological architectures, retain specific functions of the parent tumors and can faithfully recapitulate drug responses, the organoid technology fills the gap between drug screening based on classical 2D cell lines and clinical trials. Numerous studies have demonstrated that organoid can serve as an excellent model for evaluating specific responses of cancer patients [26, 69, 81, 95, 96]. Besides, it also can be an extraordinary alternative to explore the detailed causal epigenetic and genetic alterations underlying drug resistance [97]. Several organoid biobanks of cancers so far have been established for the purposes of identifying and testing novel drugs [37, 38, 98], and healthy organoids can be utilized to test toxicology.

Immunotherapy, which is among the chief novel and promising strategies, employs the patient’s own immune system to kill tumor cells. A prerequisite for immunotherapy is that malignant cells exhibit sufficient immunogenicity to trigger adequate immune response [108, 109]. Mutational status of cancer cells, which contribute to neo-antigens production, is responsible for immune responses [109, 110]. However, the intensity of immune response induced by neo-antigens of carcinoma is insufficient, which can be addressed through activating and expanding immune cells in vitro for in vivo application in patients.

Multiple studies have brought new hope for the application of organoid technology in immunotherapy, as exemplified by functional maintenance of intraepithelial lymphocytes being co-cultured with mouse intestinal organoids at the presence of interleukin-2 (IL-2), IL7, and IL-15 in the culture medium [111]. Another example is that the short-term maintenance of CD45-positive lymphocytes can be achieved through co-culture with patient-derived organoids of air-liquid interface tumors [112]. Encouragingly, co-cultures of Vδ2⁺ T lymphocytes and organoids of primary human breast epithelial have been developed successfully, and these T lymphocytes could potently eradicate triple-negative breast cancer cells [113]. These findings signify the possibility that T lymphocytes from healthy blood donors can be expanded and activated with organoids and subsequently utilized to treat patients, and the possibility that the cytotoxic effects of healthy donor-derived T cells on patient-derived tumoroids can be tested in vitro.

Personalized medicine, also called precision medicine, aims to identify effective treatment strategies for each patient through better characterization of diseases at molecular and pharmacogenomics levels. As an excellent minute incarnation of an in vivo organ, organoids are superior to conventional models, because this easily established model can better recapitulate in vivo characteristics in phenotype, genotype, and specific functions as well as physiological and pathological changes even after many generations. Organoids are endowed with enormous potential to identify the feasible optimized treatment strategy for the individual patient [29, 30, 114, 115].

Rubin MA and colleagues applied the organoid platform to identify the optimized combination therapy options for some cancer types as exemplified by uterine carcinosarcoma and endometrial adenocarcinoma harboring similar driver mutations in PIK3 catalytic subunit alpha and PTEN [29]. The uterine carcinosarcoma organoid receiving combination treatment of vorinostat and buparlisib showed strongest inhibition in comparison with other combination strategies, while the combination of buparlisib andolaparib was among the most effective strategies for the endometrial adenocarcinoma organoid [29].

Another example was that the KRAS and TP53-mutant organoid of stage IV colorectal cancer only showed notable response to trametinib, and the combination of trametinib and celecoxib was among the chief strongly effective combinational options [29]. Besides, it was also demonstrated that the novel combination of afatinib and histone deacetylase inhibitors contributed to dramatically enhanced growth suppression of colorectal tumoroids with APC mutations, even greater than the standard FOLFOX (oxaliplatin, FU and leucovorin) regimen did [29]. In addition, drug screening was also conducted on human colorectal organoids from patients, containing many cancer SCs and being resistant to 5-FU and irinotecan [116]. Organoids treated with hedgehog signal inhibitors (AY9944 and GANT61) exhibited reduced cell viability with downregulation of c-Myc, CD44, and Nanog [116], and organoids treated with the combination of AY9944 or GANT61 with 5-FU or irinotecan showed impaired cell viability in comparison to each drug alone [116]. These results reflected that inhibitors of hedgehog signaling could serve as an effective combinational candidate for the treatment of 5-FU or irinotecan-resistant colorectal tumors [116]. Based on the phenomenon that anaplastic lymphoma kinase (ALK) mutation (F1174C) promoted growth and upregulated the expression of neuroendocrine marker neuron-specific enolase in the organoids of prostate small cell carcinoma, alectinic showed more significant effects than crizotinibin terms of inhibiting ALKF1174C-expressing cell expansion [117].

Photodynamic therapy, known as a light-activated cancer therapy, supplements conventional chemotherapies and brings clinical promise for pancreatic cancer treatment [118]. As observed in organoids of metastatic pancreatic carcinoma, intelligent combination of oxaliplatin and neoadjuvant photodynamic therapy exhibited remarkably enhanced anti-tumor efficacy in comparison with any therapy alone, without augment of toxicity [118].

Although it is still in an immature stage of organoid technology in personalized medicine, further efforts can refine this model and broaden horizon in personalized medicine in replacement for conventional “one-size-fits-all” treatments.

Although organoids have a wide range of potential applications, the current version still represents a somewhat rough model, and researchers still grapple with obstacles of this technology. Firstly, organoids are imperfect reproductions. The “tissues in a dish” comprise only epithelial layer without native microenvironment including surrounding mesenchyme, immune cells, nervous system, or muscular layer [81]. Possible solutions to this limitation are to further refine organotypic culture system or to co-culture with additional cellular elements such as immune cells, stromal cells, or neural cells, as exemplified by iPSC-derived intestinal organoids containing a functional nervous system [119] and co-culture of PDAC organoids with mouse pancreatic stellate cells which differentiated into cancer-related fibroblasts [120]. In spite of these encouraging findings, an immune microenvironment around a tumor is difficult to be modeled. Immune niche of tumors is a complicated system composed of diverse immune cells including cytotoxic lymphocytes, tumor infiltrating dendritic cells, regulatory T cells, tumor-associated macrophage, and myeloid-derived suppressor cells, and tumor immune microenvironment is in dynamic changes, and there may be differences between different tumor types as well as individual patients. Secondly, fully maturation is an obstacle required to be tackled, which might affect the therapeutic potential. Thirdly, some organoid lines still cannot be expanded for long term, which could be disposed through improvement of culture medium. Fourthly, cancer organoids tend to grow more slowly than corresponding organoids from normal epithelial, thus probably contributing to the outgrowth of tumor organoids by contaminating normal epithelial cells. This problem might be addressed through improving the tissue extraction process to minimize the contaminating normal cells. Fifthly, current organoids are majorly derived from epithelium, and further investigation of cultures of non-epithelial organoids is needed, taking the recent advances in establishment of organoids induced from primary glioblastoma as an example. Lastly, the growth factors or small molecular inhibitors in culture medium may have significant effects on gene expression and signaling pathways in organoids, and may affect drug sensitivity. Further efforts are in need for addressing this problem.

In spite of these limitations, the exciting and promising organoid technology holds enormous potential to more accurately model human tumors. Up to now, highly efficient establishment of organoids has been achieved from both normal and malignant tissues. Using these amazing 3D cultures, both drug screening and personalized medicine can be prompted dramatically to better predict drug responses and guide optimized therapy strategies for an individual patient. Future efforts will doubtless bring this novel technique closer to clinical practice.